A Study on Development Methodology of Sustainable Solid

Waste Management System by Using Multi-Objective Decision

Making Model – A case study in Hoi An City, Vietnam

September 2017

HOANG MINH GIANG

Graduate School of Environmental and Life Science

(Doctor’s Course)
Okayama University
 Dissertation submitted to
Graduate School of Environmental and Life Science
of
Okayama University
for partial fulfillment of the requirements
for the degree of
Doctor of Philosophy

Written under the supervision of

Professor TAKESHI FUJIWARA
and co-supervised by
Professor KATSUYA KAWAMOTO
and
Professor YASUHIRO MATSUI

Okayama University, September 2017

 To Whom It May Concern

We hereby certify that this is a typical copy of the original doctor thesis of
Mr. Hoang Minh Giang

Signature of Seal of
the Supervisor

Graduate School of
Professor Takeshi Fujiwara Environmental and Life Science
 ABSTRACT

Solid waste generation is the result of human activities in their production and consumption
cycle. Rapid development in Vietnam causes significant increases in municipal solid waste
(MSW) generation and negative impacts on the Environment and human health due to the
inappropriate waste management system. A sustainable waste management system which
ensures environmental safety and human health becomes a critical target of Viet Nam. However,
hardly any study provides holistic methods for planning a sustainable waste management that is
environmentally effective, economically affordable and socially acceptable in socioeconomic
conditions of developing countries. Also, study on methods for evaluating social acceptance for
waste management still lagged behind environmental and economic objectives.

Thus, this dissertation aims at developing a methodology for planning a sustainable waste
management system in the social-economic conditions of Vietnam including following main
contents: (1) Development of useful sampling procedure and appropriate statistical analysis in
waste characterization. The method is readily applicable to medium and small cities in Vietnam
for waste characterization study. (2) Identification of relevant factors influencing on household
waste generation as well as prediction of MSW generation using multivariate linear regression
model (3) Development of Multi-objective Decision-making (MODM) models to identify the
optimal solutions in the selection of MSW system. Decision variables as results of the model
show the efficient waste flows, and appropriate treatment technologies to satisfy all objectives.
(4) Development of an approach to optimize the social acceptance objective in MODM
techniques by using Consensus Analysis Model (CAM) and Reference Point Method (RPM).
Those help to identify compromise solution over questionnaire survey and direct discussion with
related decision makers including authorities, stakeholders and citizen representatives.

To apply these methods to actual targets, a case study was conducted in Hoi An, a well-
known tourist city in Vietnam. Firstly, a sampling of municipal waste and analysis of
composition and generation were carried out in 2015. To determine the waste characteristics
from various sources of the city, not only waste from 321 households but also waste from tourist
sources including 9 hotels (HT), 6 restaurants (RBC), and 3 streets (STR) of tourist corners were

i
chosen. Waste collected was classified into 10 physical categories and 18 subcategories; then, 37
samples were brought to Japan for chemical analysis. The result showed that the daily per capita
household waste generation was 0.223 kg capita -1 day-1. People living in rural area generated
about a half of the amount of daily waste produced by residents in the urban area. Hotels
generated about 0.6 kg room-1 day-1 and one restaurant in HAC produced 26.18 kg day-1 in
average. Waste generation from tourist streets was 6.99 kg 100m -1 day-1 in average. The
composition of municipal waste in the city had food waste as the largest proportion (42%) and
hazardous waste as the smallest contributor (less than 1%). Total biodegradable waste (food and
yard trimmings) was approximately 53%, and combustible waste was the second significant
component of about 16% while other recyclable contributed about 20% of the municipal waste
composition.

Secondly, a questionnaire survey of households (during the period of waste characterization

sampling) was conducted to address the relevant factors regarding household waste generation.
The Bayesian model average method was used to identify factors significantly associated with
waste generation. Multivariate linear regression analysis was then applied to evaluate the impacts
of significant factors on household waste production. The model obtained from this study
indicated that household location, household size, house area per person, and family economic
activity are important determinants of the waste generation rate. The models could explain about
34% of the variation of the per capita daily waste generation rate. Diagnostic tests and model
validation results showed that the regression model could provide reliable results of estimated
household waste. The study revealed that per capita urban household waste generation is 70–
80% higher compared to a rural household. That provides a reliable information of waste
generation in different areas for better calculation and arrangement of vehicles, labours,
collection routes. The models also showed that if a family ran a business from home, the
household waste generation rate would increase by about 35%. Thus, the waste fee for business
sector might be a major factor that local authorities should take into account in waste collection
and management planning. Two other significant variables (family size and house area per
capita) do not contribute much (less than 20%) to waste generation. Variables accounting for
household income, the presence of a garden, number of rooms in a house, and percentage of
members of different ages were proven to be not significant. The study provides a reliable

ii
method for estimating household waste generation, offer decision makers useful information for
waste management policy development.

Next, a single-objective model is formulated to minimise the total cost of the system in
addition to landfill waste reduction target. For sensitive analysis, various values of waste
separation efficiency were employed as constraints of the model. 12 scenarios were developed
using combinations of 3 waste-to-landfill targets (LT: 50%, 25% and 10%) and 4 values of waste
separation rate (SE: 100%, 90%, 80%, 70%). The model identified optimal waste flows, and
proper treatment options to ensure minimum cost for each scenario. As a result, they were
evaluated: the influence of landfill reduction policies and waste separation efficiency to the cost,
pollutants emissions ( including CH 4, CO2, SO2, NOx, Heavy Metal, VOC, N2O emission) and
GHG emission of the system. It is infeasible to reach 10% of waste to landfill with the separation
rate of 70% (Scenario SC12). It means that to achieve the landfill reduction target to 10% as
expected, the waste separation efficiency of the city should be higher than 80%. System cost is
proportional with the landfill target, the higher landfill reduction target, the higher cost for the
system. Also, the comparison among a group of scenarios with the same landfill target (SC1-SC4
(LT:50%), SC5-SC8 (LT:25%) and SC9-SC11(LT:10%)) shows that the separation efficiency
affects significantly to the system cost. The lower rate of waste sorting at source, the higher the
system cost. The Scenario SC1 (LT: 50% and SE: 100%) provided the lowest cost of about 1500
US$/day and the highest GHG emission due to the complete waste separation but the amount of
waste-to-landfill is high (50% of total MSW being landfilled). The system cost and emission are
increased in SC2, SC3 and SC4 because of lower separation efficiency while the GHG emission
remains the same. Analyzing the results of SC5 to SC11, waste incineration is seemed an
effective technology to deal with the low rate of waste separation for waste-to-landfill amount
control. However, increasing of waste incineration also causes higher CO 2 emission and air
pollutants emission as the result of the analysis. Composting appears as a reliable technology for
Hoi An due to low cost and low emission. The percentage of 25% waste to landfill is an
achievable target based on the current condition of Hoi An city. Also, the model suggested that
the cost is about 1800 US$ per day and the pollutant emission is 40 tonnes per day with SC5(LT:
25% and SE: 100%) in a combination of incineration, composting and landfill as main treatment
technologies.

iii
 Then, a multi-objective optimization model was developed for identifying the optimal
solution in selecting a sustainable waste system. Firstly, a face-to-face interview survey was
conducted in 2016 with 18 local experts including authorities, stakeholders, waste managers,
scientists and citizens to draw out the appropriate waste treatment technologies and priorities of
waste management of the city. The result of the survey was analysed by consensus analysis
model (CAM) for choosing treatment options and objective functions in optimisation models.
The consensus result (CR) of Minimising Cost (CR=0.722), Minimising Emission (CR=0.512)
and Minimizing Landfill (CR=0.931) were higher than other objectives such as maximising
benefit (CR=0.222) or minimising GHG emission (CR=0.271), etc. Also, the degree of
consensus (DC) equalled to 0.808, 0.876 and 0.927 for Cost, Emission and Landfill respectively,
which means that there was significant consensus for the objectives of the system among local
experts. Also, seven waste treatment options were chosen by local experts as potential treatment
alternatives for Hoi An city as the results of CAM. Thus, the model proposed accounts for the
above three objectives including minimising cost, minimising emission and minimising waste to
landfill with chosen treatment alternatives to achieve the goal of sustainability. An optimisation
model proposes proposing the efficient waste-flow-allocation and the capacity of disposal
facilities as decision variables. The reference point method (RPM) is applied for solving the
model to get an optimal solution. Reference points (RP) are chosen by decision makers (DM)
who present for different groups involving in waste management such as civil authority,
stakeholder and residents throughout a meeting for decision making. The final solution will
determine the optimal decision variables and values of objectives which can satisfy all involving
DMs. An experiment of the decision-making process was conducted. Three colleagues were
invited to play the role of three DMs representing the city authority, the waste management
company and citizen in making decisions. Finally, a compromise solution was determined as a
result of decision-making which includes intense discussion and agreement between DMs. As a
consequence, a Decision Support System using Multi-Objective Decision-Making model was
developed which support a variety of policy makers in making decisions. The method provides
direct results of the proposed system and also the graphical visibility of chosen solution helping
the DMs to discuss adjustment to the final goal. As a result, it takes a shorter time to converge to
the compromise decision. The solution indicates that as a cost of about 2183 US$ per day, waste

iv
incinerator (no energy recovery) and anaerobic digestion combined with composting plants
should be applied associated with informal recycling activities and home treatment for
degradable waste. The amount of waste to landfill was 20 tonnes per day ( about 29.4%) and
pollution emission of 35 metric tons per day and GHG emission of 0.38 thousand tonnes CO 2-eq
per day.

In conclusion, the dissertation could propose the methodology from waste sampling and
characterization until decision-making process by using Multi-objective decision-making model
through identification of a statistical model of waste generation. The study proposed a systematic
SWM planning process supporting various decision makers including authority, business sector
and residents working together to obtain compromise solution in decision-making. Also, the
system model developed is reliable to apply in Viet Nam cities. In Viet Nam, such process and
system are not implemented yet, and a Decision Support System which considers and optimize
social acceptance is firstly proposed. Before drastically increasing of waste generation, designing
sustainable SWM system which is suitable for the local condition is inevitable. The author
proposed an effective approach to realise this situation in Viet Nam. The model and decision
support method will be consistently improved for application in the decision-making process of
waste management in different city scales in Viet Nam for our future study.

Key words: Decision-making process, Interactive method, MODM model, Optimization

model, Sustainable waste management, Waste characterization, Waste generation model.

v
TABLE OF CONTENT

ABSTRACT............................................................................................................................. i

TABLE OF CONTENT ........................................................................................................ 1

ABBREVIATIONS................................................................................................................ 5

LISTS OF TABLES............................................................................................................... 6

LIST OF FIGURES ............................................................................................................... 8

1. INTRODUCTION....................................................................................................... 11

1.1 Background ........................................................................................................... 11

1.1.1 Problem Statements .......................................................................................... 11
1.1.2 Current status of Municipal waste management in Viet Nam .......................... 14

1.2 Research objectives............................................................................................... 23

1.3 Scope of study....................................................................................................... 24

1.4 Outline of Research............................................................................................... 25

1.5 References............................................................................................................. 26

2. LITERATURE REVIEW .......................................................................................... 29

2.1 Sustainable waste management planning.............................................................. 29

2.1.1 Sustainable waste management system............................................................. 29
2.1.2 System analysis approach to Waste Management ............................................ 30
2.1.3 Multi criteria decision making models ............................................................. 31

2.2 MSW characterization .......................................................................................... 32

2.2.1 Waste composition analysis .............................................................................. 32
2.2.2 Prognosis waste generation modelling ............................................................. 35

2.3 MSW Optimization modeling............................................................................... 38

1
 2.3.1 Optimization models ......................................................................................... 38
2.3.2 Solving approaches ........................................................................................... 39
2.3.3 Gaps in sustainable waste management optimization models .......................... 40

2.4 Conclusions........................................................................................................... 41

2.5 References............................................................................................................. 42

3. METHODOLOGY AND PROPOSED CASE STUDY........................................... 48

3.1 Methodology ......................................................................................................... 48

3.1.1 Waste generation and composition study ......................................................... 48
3.1.2 Mathematical modelling of waste generation................................................... 49
3.1.3 Single optimization model for evaluation of alternative waste management
initiatives 51
3.1.4 Multi-objective optimization model for MODM ............................................... 52
3.1.5 Statistical analysis and computer software....................................................... 55

3.2 Case study in Hoi An, Viet Nam .......................................................................... 55

3.2.1 General Introduction ........................................................................................ 55
3.2.2 Current status of MSW management ................................................................ 57
3.2.3 Challenges of current solid waste management ............................................... 63

3.3 Research proposed for the case study ................................................................... 65

3.4 Reference .............................................................................................................. 67

4. MUNICIPAL SOLID WASTE CHARACTERIZATION ...................................... 70

4.1 Introduction........................................................................................................... 70

4.2 Methodology ......................................................................................................... 71

4.2.1 Research target areas and number of samples ................................................. 71
4.2.2 Sampling procedures ........................................................................................ 73
4.2.3 Physical and chemical analysis ........................................................................ 76
4.2.4 Statistical analysis ............................................................................................ 76

2
 4.3 Result and discussion ............................................................................................ 78
4.3.1 Municipal waste generation.............................................................................. 78
4.3.2 Municipal waste composition ........................................................................... 83
4.3.3 Waste characteristics ........................................................................................ 87

4.4 Conclusion ............................................................................................................ 90

4.5 Reference .............................................................................................................. 91

5. MODELLING WASTE GENERATION PROGNOSIS MODEL........................ 95

5.1 Introduction........................................................................................................... 95

5.2 Mathematical modelling of Waste generation ...................................................... 96

5.2.1 Data collection.................................................................................................. 96
5.2.2 Variables used in modelling.............................................................................. 97
5.2.3 Selection of determinant variables.................................................................... 98
5.2.4 Multivariate linear regression model ............................................................. 100
5.2.5 Testing model assumptions ............................................................................. 101
5.2.6 Model evaluation and validation .................................................................... 102

5.3 Result and discussion .......................................................................................... 103

5.3.1 Significant independent variables and selected models.................................. 103
5.3.2 Multivariate linear regression models for household waste generation ........ 107
5.3.3 Models analysis............................................................................................... 110

5.4 Conclusion .......................................................................................................... 114

5.5 Reference ............................................................................................................ 115

6. MODM Model ........................................................................................................... 122

6.1 Introduction......................................................................................................... 122

6.2 Assessment to the desires of society................................................................... 123

6.2.1 Methodologies................................................................................................. 124
6.2.2 Results of the expert's survey .......................................................................... 127

3
 6.3 Modeling waste management system ................................................................. 128
6.3.1 Waste flow analysis......................................................................................... 128
6.3.2 Waste flow modelling ...................................................................................... 130

6.4 Single objective optimization model .................................................................. 137

6.4.1 Objective function ........................................................................................... 137
6.4.2 Scenarios development.................................................................................... 138
6.4.3 Results of Scenarios analysis by single optimization model ........................... 139

6.5 Multi-objectives optimization decision-making model ...................................... 145

6.5.1 Objective functions of the model ..................................................................... 145
6.5.2 Decision-making process approach................................................................ 146
6.5.3 Results of MODM ........................................................................................... 148

6.6 Conclusion .......................................................................................................... 156

6.7 Reference ............................................................................................................ 157

7. CONCLUSION AND RECOMMENDATION ...................................................... 160

7.1 Summary of key findings.................................................................................... 160

7.2 Limitations of the study and recommendations for future studies...................... 166

ACKNOWLEDGEMENT ................................................................................................ 168

4
 ABBREVIATIONS

BMA Bayesian model averaging

BIC Bayesian information criterion
CBA Cost – Benefit Analysis
CH4 Methane
CO2 Carbone dioxide
DSS Decision support system
EIA Environmental Impact Analysis
EU European Union
Eq. Equation
GDP Gross domestic product
GHG Greenhouse gas
HAC Hoi An city
MAE Mean absolute error
MFA Material Flow Analysis
m2 Cubic meter
MSW Municipal solid waste
MODM Multi-objective decision making
NRMSE Normalized root mean square error
MSWM Municipal solid waste management
OECD Organization for Economic Co-operation and Development
P-value Probability value
% Percentage
R2 Coefficient of determination
RMSE Root mean square error
SWM Solid Waste Management

5
LISTS OF TABLES

Table 1.1. Waste generation in urban area ............................................................................ 15

Table 1.2. Waste composition and characteristics ................................................................. 16

Table 1.3. Collection rate in different urban in Vietnam....................................................... 19

Table 3.1. Waste collection schedule for Hoi An city........................................................... 60

Table 4.1. Numbers of statistical samples adopted in the survey program. .......................... 73

Table 4.2 Categories of waste separation in samples ............................................................ 75

Table 4.3 Means of waste generation and 95% confident interval. ....................................... 78

Table 4.4 Results of variance analysis................................................................................... 80

Table 4.5 Waste composition of different sources regarding tourism waste......................... 85

Table 4.6 Waste characterization in Hoi An city................................................................... 87

Table 5.1 Types of variables in linear regression .................................................................. 98

Table 5.2 Best models selected by Bayesian Model Average ............................................. 107

Table 5.3 Results of model validation ................................................................................. 113

Table 6.1 Experts and stakeholder groups being interviewed ............................................. 123

Table 6.2 Degree of compromise for consensus with different number of stakeholders .... 126

Table 6.3 The CR value of general objectives for sustainable waste management ............. 127

Table 6.4 Consensus results and Degree of consensus for treatment options ..................... 128

Table 6.5 Sets, parameters and variables ............................................................................. 131

Table 6.6 Product and residual rate of treatment alternatives ............................................. 134

Table 6.7 Energy consumption coefficients ........................................................................ 135

Table 6.8 Emission factors  p,k .......................................................................................... 135

Table 6.9 The capital cost and variable cost of treatment facilities..................................... 136

6
Table 6.10 Scenario description .......................................................................................... 139

Table 6.11 Trade-off table ................................................................................................... 148

Table 6.12 Interactions of decision making process with Multi-objective model............... 149

Table 6.13 Waste separation rate of optimal solutions ........................................................ 154

7
LIST OF FIGURES

Figure 1.1. Common flow of solid waste and collection stakeholders in Vietnam ............... 18

Figure 3.1. The location and general map of Hoi An city ..................................................... 56

Figure 3.2. MSW generation in Hoi An ................................................................................ 57

Figure 3.3. Daily waste generation rate in Hoi An city (recorded in 2014) .......................... 58

Figure 3.4. Kerbside collection in the centre of Hoi An........................................................ 59

Figure 3.5. Current MSW management in Hoi An................................................................ 60

Figure 3.6. Targets and contents of waste data collection ..................................................... 67

Figure 4.1. Target research areas. .......................................................................................... 72

Figure 4.2. Sampling ratio ..................................................................................................... 72

Figure 4.3. Waste sampling procedure .................................................................................. 74

Figure 4.4 Physical and Chemical analysis ........................................................................... 76

Figure 4.5 Boxplot of waste generation rate......................................................................... 79

Figure 4.6 Density of waste generation rate .......................................................................... 79

Figure 4.7 95% family-wise confidence level ....................................................................... 80

Figure 4.8 Boxplot of waste generation from tourism sector ................................................ 82

Figure 4.9 Correlation of mean of waste generation and number of hotel room................... 82

Figure 4.10 Waste composition of household and tourism sector. ........................................ 84

Figure 4.11 Household waste compositions of thee strata. ................................................... 84

Figure 4.12 Municipal waste compositions ........................................................................... 85

Figure 4.13 Hoi An MSW composition 18 sub-categories.................................................... 86

Figure 4.14 Physical and chemical characteristics of MSW in Hoi An city ......................... 87

Figure 4.15 Calorific value of MSW in Hoi An city ............................................................. 88

8
 Figure 5.1 Correlation coefficients among variables........................................................... 104

Figure 5.2 Predictors being chosen for the most reliable models by BMA......................... 106

Figure 5.3 Fitted line plot of model with four regressors Xplc, Xsiz, Xpa, Xbus ............... 109

Figure 5.4 Fitted line plot of model with three regressors Xplc , Xsiz, Xbus ..................... 110

Figure 5.5 Tests for linear assumption of two models......................................................... 111

Figure 5.6 The relative importance of regressors for waste generation rate ....................... 112

Figure 5.7 Observations have high influence to the model ................................................. 112

Figure 6.1 The degree of compromise for consensus (DC value) ....................................... 128

Figure 6.2 The waste flow structure of the model ............................................................... 130

Figure 6.3 Total Cost and Emission fo 12 Scenarios........................................................... 139

Figure 6.4 Treatment options for 12 Scenarios ................................................................... 141

Figure 6.5 Emission of CH4, CO2, NOx, VOC, SO2, N2O and heavy metals ...................... 142

Figure 6.6 GHG emission of alternative Scenarios ............................................................. 142

Figure 6.7 Material Flow of Waste Management System Landfill target of 25% and
separation efficiency of 100% .................................................................................................... 144

Figure 6.8 Decision making process.................................................................................... 147

Figure 6.9 MOMD model with all reference points ............................................................ 150

Figure 6.10 Pareto-Optimal solutions of decision making process ..................................... 150

Figure 6.11 Objective values and treatment alternatives of four optimal solutions ............ 151

Figure 6.12 Pollution emission of obtained Pareto-optimal solution from DMs ................ 152

Figure 6.13 Material Flow of Sustainable Waste Management System of compromise

solution........................................................................................................................................ 153

Figure 6.14 Model behavior under uncertainty of waste separation rate ............................ 154

Figure 6.15 Comparision of different separation efficiency................................................ 155

9
Figure 6.16 Distance and objective values associated with waste separation rate .............. 155

10
1. INTRODUCTION

1.1 Background

1.1.1 Problem Statements

Healthy municipal solid waste management (MSW) is a pressing issue worldwide. Annually,
global MSW generation is approximately 1.3 billion tonnes, and it is forecasted to increase to
approximately 2.2 billion tonnes in 2025 (Hoornweg & Bhada-Tata, 2012). MSW generation is
affected by economic development, the degree of industrialisation, residents’ habits, and cultures.
Income level and rate of urbanisation are highly correlated. As income and living standard
increase, consumption of goods and services correspondingly increases, as the amount of waste
produced. Therefore, the amount of MSW generated is expected to rise steeply in the next
decades. Much of the increase coming in fast-growing cities in developing countries, will be a
threat to the environment, public health and safety, so are the financial and social ramifications.

The threat of waste is an urgent issue in less developed nations. Uncollected waste due to
inappropriate waste collection and management is typically heaviest near less affluent
neighbourhoods and slums. The frequency of illness such as diarrhoea and acute respiratory
infection, linked to water pollution and the open burning of waste respectively is much higher,
especially in lower developed areas.

MSW also causes a financial burden for municipalities. In general, solid waste management
is given a very low priority in developing countries, except perhaps in the capital and large cities.
Normally, cities in developing countries spend 20% to 50% of their budgets for dealing with
waste management, and 90% of the annual budget provided for solid waste management was
probably used up within the first six months in a developing city (Ogawa, 2008). According to
Hoornweg and Bhada-Tata (2012), the cost of waste management will increase 3-4 times in
developing countries from about 20 billion US$ in 2010 to approximately 80 billion US$ in 2025.
The rate of cost increase is higher in lower developed countries.

In addition, Climate change has become a matter of public concern, the concentration of
CO2 and Methane (CH4) were increased 35% and 100%, respectively (ISWA, 2010). According
to the report of IPCC (2007), the amount of Greenhouse gas (GHG) emission from waste sector

11
accounted for about 3% of total artificial GHG emission, in which 90% is methane gas (Scheutz,
Kjeldsen, & Gentil, 2009). In term of Methane emission, Waste sector only contributed around
18% globally (Bogner, 2007), mainly from landfill and wastewater treatment facilities. However,
MSW management is a small contributor to GHG emission, but it is becoming a significant GHG
mitigator due to its potential for material and energy recovery (UNEP, 2010).

Thus, a sustainable MSW management system has become crucial for the sustainable
development strategy of developing countries. To develop a sustainable solid waste management
system, a variety of procedures should be conducted efficiently including sampling, surveying to
address the problem and obtain waste information, modelling and simulation the reality,
calculation economic and environmental impacts of potential management options,
implementation of the decision-making process. However, the lack of attention in the study on
how to develop a sustainable waste management has led to the improper waste management
system in developing countries. The relative importance of procedures which play a major role in
waste management planning are explained in the following.

Firstly, planning, design, and operation of a sustainable MSW management system require
the understanding of the features the waste stream (Abu Qdais, Hamoda, & Newham, 1997;
Chang & Pires, 2015b). For integrated waste planning, accurate and reliable data on the waste
composition and generation are needed for evaluation of optimal treatment options. However,
physical, geographical, sociocultural, economic, and political factors may have influences on the
composition and generation of municipal solid waste (MSW) (Gallardo, Carlos, Peris, &
Colomer, 2014; Gidarakos, Havas, & Ntzamilis, 2006). Especially, economic development may
lead to significant influences to the environment on resource consumption as well as pollution
generation. For instance, one of the major impacts of tourist development is the change in
generation rate and composition of MSW as well as household waste. Some previous studies
have reported the increase in MSW in the tourist areas resulting from the high number of tourists
during the tourist season (Denafas et al., 2014; Espinosa Lloréns et al., 2008; Shamshiry et al.,
2011; Teh & Cabanban, 2007). Thus, waste composition and waste generation rates can be
different in places influenced by many relevant factors mentioned above. However, studies on
waste have still lagged behind where it concerns waste arising from developing countries due to

12
the near absence of reliable data on MSW management and lack of attention from authorities and
waste managers (Ezeah, Fazakerley, & Byrne, 2015).

Secondly, the prediction of MSW generation also plays a major role in a solid waste
management planning. Apparently, predicting waste generation is increasingly essential in waste
collection planning, waste treatment strategies and establishing waste policies toward a
sustainable waste management system (Abbasi, Abduli, Omidvar, & Baghvand, 2012; Chen &
Chang, 2000; Thanh & Matsui, 2011). In developing world, one of the most challenges faced by
local governments is the prognosis of solid waste quantities to have appropriate actions and plan
(Ghinea et al., 2016). For instance, in Viet Nam, the National Technical Regulation QCVN
07:2010/BXD (MOC, 2010) provided a method to estimate waste generation for five urban types
based on population and waste generation rate determined in the document. However, the results
of prediction are not reliable in term of practical application for different cities because the solid
waste generation is impacted not only by demographic factor but also by social, economic as
well as other factors (e.g. family expense or waste prevention policies). Therefore, the later
edition of this regulation, QCVN 07:2016 (MOC, 2016) uses neither this method for waste
generation estimation nor any other model instead. Lack of research and method on waste
generation estimation has led to a considerable challenge in municipal waste management in
developing countries.

Then, the decision making process normally have been carried out by authority alone and
lack of taking consideration of the opinions of involved parties such as business sectors, citizens,
NGO etc. Thus, the social acceptance, which is a crucial aspects of sustainability has not been
well presented in waste management plan. Different stakeholders normally require different
interest from waste management system. For instance, authority might requires a low cost sytem
with minimum of waste to landfill. Residents are interesting in a system with fast collection and
less pollution. Meanwhile, business sectors care more about how to maximize the benefit or
minimize the cost. Waste management system always has conflicts needs to be solved by all
participants. Thus, a requirement of development a Decision Support System (DSS) helping a
variety of decision makers from various stakeholder group participating in decision making
process is essential. Especially, the decision making process needs the involvement of citizens

13
(people who generated waste and being directly affected from waste management system) in
full-fill the meaning of sustainability.

Lastly, SWM is a significant issue in sustainable development encompassing technical,

socio-economic, legal, ecological, financial, political and cultural components (Chang & Pires,
2015a). System analysis which provides unique, interdisciplinary support for strategic analysis
and decision-making procedures, has been lagged behind in developing countries due to the lack
of studies and knowledge of implementing system analysis in waste management. System
analysis approach have been applied to analyse for SWM over the last few decades (Pires,
Martinho, & Chang, 2011). Multi-Objective Decision Making approach has been developed to
solve the conflict of multi-objective problems and balancing the environmental and economic
goals of the system. However, previous studies in MODM models are well addressed by
presenting a decision variable for the selection of technologies rather than environmental impacts.
The environmental impact should consider various of pollutants emission from variety of
treatment technologies, not only take into account for GHG emission or CO2 emission or only
emission from incinerator, landfill. Moreover, the waste cycle from generation source to final
disposal was well calculated in LCA studies but not well presented in previous MODM studies.
Thus, a MODM model which can fill those gaps mentioned above is crucial for developing
countries.

1.1.2 Current status of Municipal waste management in Viet Nam

Vietnam is an S-shaped country located in the Center of South-East Asia which has 3,730
km mainland border with China in the North, Laos and Cambodia in the West. The total land
area of 330,967 km2 with a population of approximately 92 million in 2015 (GSOVN, 2016).
Vietnam is developing rapidly and undergoing urbanisation with current GDP of about 193.6
billion US$.

Solid waste generation is the result of human activities in their production and consumption
cycle. Rapid urbanisation and industrialisation in Vietnam have led to thousands of tonnes of
municipal solid waste (MSW) generated daily. Currently, solid waste generation is assessed to be
more than 24 million tonnes per year, with a likelihood of reaching 52 million tonnes by the year
2020 (T. K. T. Nguyen, 2014). Increasing MSW generation has been becoming an emerging

14
environmental issue for authorities in Viet Nam (D. L. Nguyen, Hoang, & Bui, 2013). The
growing waste amount causes negative impacts on the environment and human health due to the
inadequate disposal of waste (Ngoc & Schnitzer, 2009). Also, 80% of MSW was disposed of in
landfills without being recycled, refelecting the material and energy losses of the society (Ghinea
et al., 2016). Thus, Integrated waste management has become significantly important regarding
recycling material and energy from MSW as well as resource conversations (van de Klundert,
Anschütz, & Scheinberg, 2001; Zurbrugg, Gfrerer, Ashadi, Brenner, & Kuper, 2012).

1.1.2.1 . Waste Management in Vietnam

Waste generation and composition

Table 1.1 presents the significant increase in waste generation rate and the total quantity of
MSW during the period of 2007-2010 and prediction for 2020 and 2025. Annually, the average
increasing rate of MSW in urban areas is 10-16% per year, and the waste generation per capita
rate is much higher in the major cities such as Ha Noi, Ho Chi Minh, Da Nang. The rapid of
urbanization and economic growth, as well as the increase in living standards and the changing
life styles, has led to the increase in municipal waste generation, especially in urban areas.

The per capita waste generation levels generally increase, in correlation to the improvement
in the standard of living. According to the data reported by provinces, the average daily
generation rates in kg/person/day range from 0.8 to 1.2 kg/person.day in the major cities and
from 0.35 to 0.5 kg/person.day in small towns (T. K. T. Nguyen, 2014). On average, the
consumed amounts of energy, goods, and food of urban residents are about 2-3 times higher than
those of rural residents in Vietnam. Thus, urban dwellers produce about twice as much waste as
their rural counterparts.

Table 1.1. Waste generation in urban area

Content Unit Year

2007 2008 2009 2010 2020 2025
Urban Population million 23.8 27.7 25.5 26.22 44 52
Percentage of urban citizens % 28.2 28.99 29.74 30.2 45 50

15
Waste generation rate in urban kg/capita.day 0.75 0.85 0.95 1.0 1.4 1.6
Total MSW generation ton/day 17.68 20.85 24.22 26.22 61.6 83.2

Source: (MONRE, 2011)

The composition of MSW in Viet Nam is diverse. It mainly consists of a large organic
fraction (56-77%), followed by recyclable waste (10-14%) and paper (2-7%). The MSW contains
recyclable materials (paper, plastic, glass, metals, etc.), hazardous waste (paints, pesticides, used
batteries), and degradable materials (fruit and vegetable peels, food waste). Table 1.2 shows the
composition and waste characteristics in some cities in Viet Nam. This suggests that there is a
tremendous potential for the implementation of biological processing and recycling activities for
MSW in Vietnam. Presently the solid waste generation is assessed to be more than 15 million
tons per year with approximately 80 % from municipal sources, 17 % from industrial sources and
the remaining 3 % from other sources. By the year 2010, the expected solid waste generation is
24 million tons per year, with a likelihood of reaching 52 million tons by the year 2020 (D. L.
Nguyen et al., 2013).

Table 1.2. Waste composition and characteristics

Composition Unit City

Hanoi Da Nang Hue Pleiku
Organic % 53.8 66 55 60.49
Plastic % 3.42 4 5.2 12.7
Paper, carton % 4.2 3.1 4.4 9.65
Metal % 1.4 4.9 7 1.16
Glass % 1 0.9 1.8 0.13
Inert % 28.18 16.4 23 12.6
Rubber % 4.9 1.6 1.5 2.8
Textile % 1.7 2.3 3 0
Hazardous % 1.4 0.8 0.8 0.4
Moisture content % 43.04 51.2 50 50.5
Ash content % 13.7 16 15.5 13.9
Bulk density ton/m3 0.41 0.4 0.4 0.38

16
 Recyclable material % 16.62 16.8 22.9 26.51

Source: (T. K. T. Nguyen, 2014)

Waste collection and transportation

In Viet Nam, private companies and informal recyclers participate partly in MSW
management. On the other hand, the Urban Environment Company (URENCO) of the city has
the highest responsibility to collect, transport and disposal the waste generated in the areas
including residential waste, street waste, and waste from commercial areas, offices, markets,
industrial zones, hospital, etc. MSW generation from different sources are normally collected to
be stored temporarily in collection points. Then, it will be transported to disposal sites by
transportation vehicles such as compression truck. The systematic diagram of waste collection
and transportation is presented in Figure 1.2.

The average percentage of solid waste collection is about 72 % for the whole country, of
which the collection rate in urban areas is increasing from 80–82 % (2008) to 83–85 % (2010)
and in the countryside about 40–45 % (MONRE, 2011; T. K. T. Nguyen, 2014). The rates of
collection efficiency in some Vietnamese cities are presented in Table 1.3. Open burning and
illegal dumping are popular, especially in the out-of-service areas. Waste generated is dumped in
garden area, by the roadside, in ditch or lake also is burnt in the area adjacent to properties or at
roadside by resident and commercial waste generator. Solid waste from households is collected
by handcart or waste collection vehicle running through streets according to a planned schedule,
then gathering to planned collection points. In the area where the waste collection service is
provided, waste is dumped in the street without any containment. It can be blown by wind or
washed into the drain or ditch of the city drainage system by rainfall, thereby contributing to
littering in the city and surface water pollution, respectively.

The collection time is from 10.00 p.m. to 6.00 am in order to avoid working at the high
temperatures in daytime, public complaint and traffic congestion. The inapproplicate collection
causes odor, air, water pollution due to waste on the streets, being blown by wind, washed into
the drainage system by rainfall.

17
 Landfill is the most popular method for MSW disposal in Vietnam at present. There are
totally 98 landfills in operation in the whole country with 16 sanitary landfill sites, and the
remaining are open dumping sites and unsanitary landfills. Thus, 76-82% of the total collected
MSW quantity is disposed at open-dumped sites and landfills (MONRE, 2011). MSW is directly
disposed of in uncontrolled and poorly managed manners such as no leachate collection system,
poor design of bottom layer, no daily cover layer, and no landfill gas collection equipment, those
of which lead to serious environmental problems and public health threat. In addition, illegal
dumping and disposal of waste to rivers, lakes, oceans, drainage channels, empty lots and
roadsides due to inefficient collection and transportation system is also a serious problem in Viet
Nam.

SOURCES COLLECTION AND TRANSPORTATION TREATMENT AND

DISPOSAL

URENCO’s responsibility
Handcarts,
HOUSEHOLD 3 wheels bike Collection points COMPOSTING
or waste storage <10%
of facilities,
STREET, PUBLIC LANDFILL
buildings
FACILITIES 76-82%

Transported by
MARKETS,
INCINERATION
specialized struck
COMMERCIAL, <1%
3 wheels bike
INSTITUTE
Recycled, treated by
Transported by
private companies,
specialized struck
HOSPITALS, craft villages
3 wheels bike
HEALTHCARE
8-12%
Collected by waste

INDUSTRY pickers, scavengers, Normal waste

junk buyers Hazardous waste

Recyclable waste
Private companies, individuals,
informal activities
Sources: (T. K. T. Nguyen, 2014; Thanh, 2010)

Figure 1.1. Common flow of solid waste and collection stakeholders in Vietnam

18
 Composting is a useful recycling form of organic waste to produce a clean soil conditioner.
It can increase the rate of material recovery, and it is known as a cost-effective means for
treatment of MSW. However, the proportion of MSW to be composted is not really high in
Vietnam. The summary of the capacity of all MSW composting plants is still less than 2,500
tonnes/day which is about 10% total quantity of MSW generated (MONRE, 2011). In reality,
due to the difficulty in the selling of composting products, most of composting plants are not
operating at full capacity. That may be as the result of following reasons: no market survey
before construction of the facilities, inadequate control of the quality and quantity of the compost,
technical constraints, waste not separated at sources, etc.

Table 1.3. Collection rate in different urban in Vietnam

Type of Name Collection Type of Name Collection

Urban Rate (%) Urban Rate (%)
Special Hanoi 90 – 95 Type 3 Bac Giang 80
Urban Ho Chi Minh 90 – 97 Bao Loc 70
Type 1 Hai Phong 80-90 Bac Lieu 52
Da Nang 90 Vinh Long 75
Hue 90 Type 4 Song Cong 80
Quy Nhon 61 Tu Son 51
Type 2 Thai Nguyen 80 Go Cong 60
Viet Tri 95 Cam Ranh 90
Nam Dinh 78 Type 5 Tua Chua 75
Thanh Hoa 84 Tien Hai 74

Source: (MONRE, 2011)

Waste treatment and disposal

Municipal solid waste of Vietnam is usually of high moisture, and low calories (900–1,100
kcal/kg). Incineration requires substantial investment, operation and maintenance cost. Therefore
the application of incineration for MSW treatment is not very much practised. However, it is
applied more for treatment of the hazardous waste and the medical waste generated from
hospitals in Viet Nam. Ministry of Health estimated that incinerators treated only 37 % of total
health care waste and the remain was treated in inappropriate ways (T. K. T. Nguyen, 2014).
19
There is only one large-scale MSW incinerator with a capacity of 300 tonnes/day installed in Son
Tay town and one medium scale waste-to-energy incinerator for industrial waste in Nam Son
complex, Hanoi city.

Legal documents related to waste management

Decision No.2149/2009/QD-TTg dated December 17, 2009 by the Prime Minister on

approving the national strategy for integrated management of solid waste up to 2025, with a
vision to 2025 

Decision No.1440/QD-TTg dated October 06, 2008 of the Prime Minister approving the
Planning on construction of solid waste treatment facilities in three northern, central Vietnam
and Southern key economic regions up to 2020; 

Decision 1216/QD-TTg dated 05 September, 2012 on the National Strategy on Environment

Protection to 2020, with Visions to 2030

For Landfills Management:

Joint Circular No. 01/2001/ TTLT-BKHCNMT-BXD dated January 18, 2001 by the
Ministry of Construction and the Ministry of Science, Technology and Environment on guiding
the regulations on environmental protection for the selection of location for, the construction and
operation of, solid waste burial sites

For Power production from MSW:

Decision No. 1030/QD-TTg dated 20 July 2009 approving “Proposal for the Development
of Vietnam Environmental Industry up to the year of 2015, vision to 2025”.

For Power generation from MSW:

 Decision No. 249/QD-TTg dated 10 Feb, 2010 by the Prime Minister on approving the
"service project development environment 2020”

 Decision No. 18/2008/QD-BCT dated July 18, 2008 of the Ministry of Industry and Trade
on promulgation of regulation on avoided cost tariff and standardized electric power purchase
agreement for small renewable energy power plants

20
  Decision No. 31/2014/QD-TTg dated May 05, 2014, on supporting mechanism for
development of electric power generation projects using solid waste in Vietnam

Climate change:

 Decision No.2139/QD-TTg dated 5 Dec, 2011 by the Prime Minister on approving the
national strategy for climate change

 Decision No.130/2007/QD-TTg dated 02 Aug, 2007 by the Prime Minister on several

financial mechanism and policies applied to investment projects on clean development
mechanism

 Decision No.1775/QD-TTg dated 21 Nov, 2012 by the Prime Minister on several

financial mechanism and policies applied to investment projects on clean development
mechanism

1.1.2.2 . MSW management problem identification and solutions

One of important problems that lead to inappropriate waste management in Viet Nam is that
the local government are not adequately equipped to provide proper waste treatment and
collection service. It is due to various reasons such as lack of resources, managerial capacity as
well as financial issues. The main problems will be list as following:

Technical constraints: Waste treatment technologies such as composting, incineration,

anaerobic digestion is not applied successfully in Viet Nam due to lack of technical attention. For
instance, waste treatment technologies developed domestically are cheap but the technologies
tend to have low-quality elements or lack of technical, environmental attention. Thus they
sometimes get failed or generates much pollution in operation. The imported technologies, on the
other hand, are much more expensive, but getting no attention about the feasibility for local
waste and conditions such as waste characteristics, technical levels of operators, managerial
capacity, economic affordability, etc. As a result, many facilities have stopped working or are
working inefficiently with emitting pollution. Low implementation of waste separation at source
is also a cause of ineffective treatment. Thus, improvement of waste separation activities as well
as evaluating an adequate levels of waste separation rate is essential for sustainable waste
management planning.

21
 Institutional constraints: There seems to be a consensus that weak institutions are a major
issue in Viet Nam as well as other developing countries (Wilson, 2007). Strengthening and
capacity building should be important to develop the more efficient waste management system.
Waste policy and legal framework are required so that they should be able to set up sustainable
objectives and targets in certain time. Moreover, scientific research and applied studies which
can overcome the lack of attention should be encouraged to enhance the efficiency and capacity
of decision makers. For example, the lack of reliable waste data and information due to poor
condition of standardised methods and studies leads to uncertainty of input data for waste
management planning and calculation. Thus, standardised methodologies of waste
characterization study is essential to get reliable data for waste management planning, and so
decision making support studies will be very beneficial for creating suitable waste management
initiatives as well as applied technologies studies, which will provide more affordable and
technically feasible treatment solutions. Also, awareness education for residents and authorities
also need to get more attention and investment and within the environmental protection policy,
the ‘‘3R Initiative’’ of Reduce- Reuse-Recycle has been raised as an important issue in need of
close attention.

Financial constraints: Multiple studies on waste management in developing countries have

cited financial and institutional constraints as the main reasons for inadequate disposal of waste,
especially where local governments is weak at performing initiative or underfinanced even
though rapid population growth continues (Zhu et al., 2008). In general, solid waste
managements is given a tiny priority in developing countries. The same problem occurred in
Vietnam, where limited funds were provided to the solid waste management sector by
government and distribution of these limited funds were mismanaged due to low managerial
capacity. Thus, the level of waste treatment service that is required for protection of public health
and the environment are not attained.

Decision-making support system constraints: In Viet Nam and many other developing
countries, there is lack of decision support system being developed to help decision makers in
making waste management plan, especially citizens and other stakeholders are hardly to get
involve in decision-making process. On the other hand, there is no MODM model study was

22
conducted to optimize various conflicted objectives for sustainable waste management plan.
Thus, developing a DSS using MODM model to support sustainable waste management planning
is essential in Viet Nam towards sustainable development of the country.

1.2 Research objectives

The overall aim of this dissertation is to obtain the knowledge of methodologies to develop a
sustainable solid waste management system which will include various steps such as waste
characterization, modelling and simulation to get optimal results, and creating a framework of
working with DM to obtain the final solution of waste management. The main objective is to
achieve a waste management system associated with sustainable development in the social-
economic conditions of Vietnam which ensure following targets: environmental effectiveness,
economic affordability and social acceptance.

In this study, detail targets needed to be done as listed as follows:

(1) To obtain and analyse the waste generation and composition from various MSW sources
including household, commercial, tourists to get the reliable input data for waste management
modelling. In addition, a waste characterization method including sampling procedures and
appropriate statistical analysis will be developed and suggested as a reference that is readily
applicable to medium and small cities in Vietnam for waste characterization study.

(2) To analyse the influence of correlated factors to household waste generation by linear
regression analysis and a multivariate linear regression model is developed to estimate household
waste produced.

(4) To analyse the Waste management strategies of waste-to-landfill control and efficiency
of waste separation at source, a single objective optimization (minimizing Cost) is proposed to
compare costs, GHG emission and pollutants emission of 12 different senarios.

(3) Optimisation models to identify the optimal solutions for the sustainable waste
management system is simulated to support the decision-making process. Decision variables of
the optimisation model will show the efficient waste flows, and appropriate treatment
technologies to reach the objectives.

23
 (5) To find a optimal solution in minimizing Cost, minimizing Emission and minimizing
Waste to landfill, an mutiple objective optimization models which uses DM’s references as
input-parameter is proposed in order to achieve the best solution of waste flow allocation and
treatment options and agreement of all DMs.

(6) To create a framework of decision making process based on Reference points method by
experiencing meeting as well as working with related decision makers (MD) including
authorities, stakeholders, and citizens representatives.

1.3 Scope of study

The study will focus on MSW management in Viet Nam, the case study is conducted in Hoi
An city ,a city on the coast of the East Sea in the South Central Coast region of Vietnam, located
in Quang Nam province and it is recognized as a World Heritage Site by UNESCO. The city has
natural land area of 6,171.25 ha with the total population of the city in 2013 is around 93
thousands and the number population density of 1,508 people km2 (HASD, 2013). It comprises
13 administrative wards including an island ward. The city is a unique urban in Vietnam which
consists many types of urban including: urban and rural places and especially it is a famous
touristic city (Hoi An ancient town), that can lead to great differences of waste generation from
different places (Abu Qdais et al., 1997).

Tourism is the most dynamic industry and the biggest economic contributor of the city. In
2013, the total income of tourist service and retail sales (from restaurants only) was 1,500 and
1,100 billion VND respectively. Meanwhile, the total industrial production value of the city was
only 212 billion VND (HASD, 2013). Also, the MSW generation increases associated with the
growth of the number of tourists. Thus, tourist industry might has affected strongly to the
variation of waste generation and composition.

In general, MSW in Hoi An is currently disposed of at an open dumping landfills without

taking any operational and engineering control. Besides, one composting plant with a capacity of
55 tons per day has been operating inefficiently. The opened dump landfill and the composting
plant have caused huge adverse effects on the environment and public health. Hence, MSW
management is one of the most important issues of Hoi An and developing a sustainable
municipal solid waste management system has become a necessity in the City.
24
1.4 Outline of Research

To obtain the proposed objective of the study, the dissertation contents including 6 chapters
as followed:

Chapter 1, General Introduction to the study

Chapter 2, a literature review of methods and results of previous studies related to this work
is reviewed. The method of MSW characterization, such as monitoring of waste generation,
classification of waste composition, relevant factors of waste generation, statistical analysis,
modelling is mentioned. Moreover, the optimisation model and application of MCDM is also
reviewed in the literature review.

Chapter 3, explain the case study and methodologies applied in the research. After
introduction of the case study, the section describes the method for waste sampling, waste
characterization and statistical analysis. Then, modelling technique of optimisation model
implemented in this work is presented. Lastly, the method of the multi-objective decision-
making process to the optimal solution for waste management is explained.

Chapter 4 deals with the analysis of MSW characterization. It will account for the detail of
household waste, hotel waste, restaurant waste and street waste generation as well as
composition. Differences in waste generation and composition from different areas in the city is
also examined.

Chapter 5, this chapter describes the development of a prognosis model for solid waste
generation from households in Hoi An city, a famous tourism city in Vietnam. Bayesian model
average method (BMA) was applied to evaluate impacts of various factors on household waste
generation. Then, multivariate linear regression model is proposed for estimating waste
generation.

In Chapter 6, a single optimisation model to minimise cost of waste management is

proposed to evaluate the impacts of landfill target initiatives and separation efficiency. Then, it
presents the results of consensus analysis model of the local expert opinion for choosing
objectives and potential treatment alternatives of the system. Lastly, a multi-objective model is

25
proposed and an optimal solution is determined by an interactive method for discussion of
decision makers. The result propose a sustainable waste management plan for the city,

Chapter 7 summarizes the main findings of this dissertation. Drawbacks of this study and
recommendations for further study is also presented in this chapter.

1.5 References

Abbasi, M., Abduli, M., Omidvar, B., & Baghvand, A. (2012). Forecasting municipal solid
waste generation by hybrid support vector machine and partial least square model. International
Journal of Environmental Research, 7(1), 27-38.
Abu Qdais, H. A., Hamoda, M. F., & Newham, J. (1997). Analysis of Residential Solid Waste
At Generation Sites. Waste Management & Research, 15(4), 395-405.
doi:10.1177/0734242x9701500407
Bogner, J., Abdelrafie, A.M., Diaz, C., Faaij, A., Gao, Q., Hashimoto, S., Mareckova, K.,
Pipatti, R. & Zhang. T. (2007). Waste management. Retrieved from Cambridge, UK and New
York, NY, USA:
Chang, N.-B., & Pires, A. (2015a). Systems Engineering Tools and Methods for Solid Waste
Management Sustainable Solid Waste Management (pp. 235-299): John Wiley & Sons, Inc.
Chang, N.-B., & Pires, A. (2015b). Technology Matrix for Solid Waste Management
Sustainable Solid Waste Management (pp. 19-97): John Wiley & Sons, Inc.
Chen, H. W., & Chang, N.-B. (2000). Prediction analysis of solid waste generation based on
grey fuzzy dynamic modeling. Resources, Conservation and Recycling, 29(1-2), 1-18.
doi:10.1016/s0921-3449(99)00052-x
Denafas, G., Ruzgas, T., Martuzevičius, D., Shmarin, S., Hoffmann, M., Mykhaylenko, V., . . .
Ludwig, C. (2014). Seasonal variation of municipal solid waste generation and composition in
four East European cities. Resources, Conservation and Recycling, 89, 22-30.
doi:http://dx.doi.org/10.1016/j.resconrec.2014.06.001
Espinosa Lloréns, M. d. C., Torres, M. L., Álvarez, H., Arrechea, A. P., García, J. A., Aguirre,
S. D., & Fernández, A. (2008). Characterization of municipal solid waste from the main landfills
of Havana city. Waste Management, 28(10), 2013-2021.
doi:http://dx.doi.org/10.1016/j.wasman.2007.07.004
Ezeah, C., Fazakerley, J., & Byrne, T. (2015). Tourism Waste Management in the European
Union: Lessons Learned from Four Popular EU Tourist Destinations. American Journal of
Climate Change, 4(05), 431.
Gallardo, A., Carlos, M., Peris, M., & Colomer, F. J. (2014). Methodology to design a
municipal solid waste generation and composition map: A case study. Waste Management,
34(11), 1920-1931. doi:http://dx.doi.org/10.1016/j.wasman.2014.05.014
Ghinea, C., Drăgoi, E. N., Comăniţă, E.-D., Gavrilescu, M., Câmpean, T., Curteanu, S., &
Gavrilescu, M. (2016). Forecasting municipal solid waste generation using prognostic tools and
26
regression analysis. Journal of Environmental Management, 182, 80-93.
doi:http://dx.doi.org/10.1016/j.jenvman.2016.07.026
Gidarakos, E., Havas, G., & Ntzamilis, P. (2006). Municipal solid waste composition
determination supporting the integrated solid waste management system in the island of Crete.
Waste Management, 26(6), 668-679. doi:http://dx.doi.org/10.1016/j.wasman.2005.07.018
GSOVN. (2016). Area, population and population density by province. Population and
Employment. Retrieved from http://www.gso.gov.vn/default_en.aspx?tabid=774
HASD. (2013). Statistical Yearbook - Hoi An city. Hoi An Statistical Department, Viet Nam.
Hoornweg, D., & Bhada-Tata, P. (2012). What a Waste : A Global Review of Solid Waste
Management. Retrieved from Washington, DC:
https://openknowledge.worldbank.org/handle/10986/17388 License: CC BY 3.0 IGO.
IPCC. (2007). Climate Change 2007: Synthesis Report. Retrieved from Geneva, Switzerland, :
http://www.ipcc.ch/publications_and_data/publications_ipcc_fourth_assessment_report_synthesi
s_report.htm
ISWA. (2010). ISWA White Paper: Waste and Climate Change C. F. Gary Crawford, Jens
Aage Hansen, Antonis Mavropoulos (Ed.)
MOC. (2010). QCVN 07:2010/BXD National Technical Regulation Chapter 9: Solid waste
collection, separation, transportation, treatment system and public toilet. Hanoi, Vietnam.
MOC. (2016). QCVN 07:2016/BXD National Technical Regulation Chapter 9: Technical
Infrastructure Works, Solid Waste Treatment and Public Toilet.
MONRE. (2011). Báo cáo môi trường quốc gia 2011: Chất thải rắn (National Environment
Rerport 2011: Solid Waste). Hà Nội, Việt Nam: Bộ Tài nguyên và Môi trường.
Ngoc, U. N., & Schnitzer, H. (2009). Sustainable solutions for solid waste management in
Southeast Asian countries. Waste Manag, 29(6), 1982-1995. doi:10.1016/j.wasman.2008.08.031
Nguyen, D. L., Hoang, M. G., & Bui, X. T. (2013). Challenges for municipal solid waste
management practices in Vietnam. Waste Technology, 1(1), 17-21.
Nguyen, T. K. T. (2014). Municipal solid waste management in Vietnam challenges and
solutions Municipal Solid Waste Management in Asia and the Pacific Islands (pp. 355-377):
Springer.
Ogawa, H. (2008). Sustainable solid waste management in developing countries: waste
management. Imiesa, 33(9), 57-71.
Pires, A., Martinho, G., & Chang, N.-B. (2011). Solid waste management in European
countries: A review of systems analysis techniques. Journal of Environmental Management,
92(4), 1033-1050. doi:http://dx.doi.org/10.1016/j.jenvman.2010.11.024
Scheutz, C., Kjeldsen, P., & Gentil, E. (2009). Greenhouse gases, radiative forcing, global
warming potential and waste management — an introduction. Waste Management & Research,
27(8), 716-723.

27
 Shamshiry, E., Nadi, B., Bin Mokhtar, M., Komoo, I., Saadiah Hashim, H., & Yahaya, N.
(2011). Integrated models for solid waste management in tourism Regions: Langkawi Island,
Malaysia. J Environ Public Health, 2011.
Teh, L., & Cabanban, A. S. (2007). Planning for sustainable tourism in southern Pulau
Banggi: an assessment of biophysical conditions and their implications for future tourism
development. Journal of Environmental Management, 85(4), 999-1008.
Thanh, N. P. (2010). A Study on Eveluation Methodologies for Household Solid Waste
Management toward a Sustainable Society in Vietnam. (Doctor), Okayama University, Okayama,
Japan.
Thanh, N. P., & Matsui, Y. (2011). Municipal solid waste management in Vietnam: Status and
the strategic actions. International Journal of Environmental Research, 5(2), 285-296.
UNEP. (2010). Waste and Climate Change: Global trends and strategy framework
Osaka/Shiga: UNEP.
van de Klundert, A., Anschütz, J., & Scheinberg, A. (2001). Integrated sustainable waste
management: the concept. Tools for decision-makers. experiences from the urban waste
expertise programme (1995-2001): WASTE.
Wilson, D. C. (2007). Development drivers for waste management. Waste Management &
Research, 25(3), 198-207.
Zhu, D., Asnani, P., Zurbrugg, C., Anapolsky, S., & Mani, S. (2008). Improving municipal
solid waste management in India. World Bank, Washington, 1-8.
Zurbrugg, C., Gfrerer, M., Ashadi, H., Brenner, W., & Kuper, D. (2012). Determinants of
sustainability in solid waste management--the Gianyar Waste Recovery Project in Indonesia.
Waste Manag, 32(11), 2126-2133. doi:10.1016/j.wasman.2012.01.011

28
2. LITERATURE REVIEW

2.1 Sustainable waste management planning

2.1.1 Sustainable waste management system

Integrated waste management is defined as a combination of combines waste stream,

waste collection, recycling, treatment and disposal, in addition, other monitoring activities,
risk and pollution prevention, with the objective of achieving human health and
environmental protection. Generally, functional elements of waste management are
generation source, storage at source and separation, collection, transfer and transport,
recycling and treatment; final disposal.

The conventional waste management approach is that waste generation, collection and
disposal systems are planned as independent operations. However, all elements are very
closely interlinked and have interaction with each other. There are also interaction between
the physical components of the system and the conceptual components that include the social
and environmental spheres.

The traditional reductionist approaches of “flame, flush or fling” to waste management is

unsustainable as it lacks flexibility and long term thinking (Jeffrey K, 2010). The shift to
more sustainable society requires greater sophistication to manage waste. Solid waste
management systems need to ensure human health and safety. According to F.R. McDougall
et al. (2001), a sustainable system for solid waste management must be environmentally
effective, economically affordable and socially acceptable.

Environmental effectiveness the waste management system must reduce as much as

possible the environmental burdens of waste management (emissions to land, air and water,
such as CO2, CH4, SOx, NOx, BOD, COD and heavy metals).

Economic affordability: the waste management system must also operate at a cost
acceptable to the community, which includes all private citizens, businesses and government.
The costs of operating an effective solid waste system will depend on existing local
infrastructure, but ideally should be little or no more than existing waste management costs.

29
 Social acceptance: the waste management system must operate in a manner that is
acceptable for the majority of people in the community. This is likely to require an extensive
dialogue with many different groups to inform and educate, develop trust and gain support.

Achieving sustainable goals involves balancing social, economic and environmental

perspectives is difficult, and possible conflicts of objectives related to the three pillars of
sustainability would be inevitable. There will always need a trade-off. It is necessary to
acknowledge and deal with these conflicting objectives across domain boundaries in the
diverse spectrum of project with system thinking.

2.1.2 System analysis approach to Waste Management

Sustainable waste management is a complex tasks that requires appropriate technical

solutions, sufficient organizational capacity, managerial capacity of government as well as
agreement and cooperation of various stakeholders (Zarate et al., 2008). According to
Marshall et al. SWM in developing countries requires an approach that can handle
complexity due to the interaction and complexity between the physical components of the
system and the conceptual components that include the social and environmental spheres
(Jeffrey K, 2010). Moreover, the conventional reductionist approach divides interacting
system and their elements into smaller parts that could not tailor to handle complexity.

The system approach developed out of an attempt to unify science. A system approach
handles the complexity more efficiently, particularly at the nexus of eco-social systems, could
contribute substantially to solid waste management research and decision making in
developing countries (Jeffrey K, 2010).

System analysis has provided unique, interdisciplinary support for SWM policy analysis
and decision making over the last few decades. N.-B. Chang et al. (2011) highlighted 14 of
classified tools and methods, in two domains including system engineering methods and
system assessment tools to illustrate the SWM challenges, trends and perspectives.

(1) The system engineering methods includes cost-benefit analysis (CBA) or benefit-
cost analysis (BCA), forecasting model (FM), simulation model (SM), optimization model
(OM), Multi-criterion decision making (MCDM) and intergrated modeling system (IMS).

(2) The system assessment tools consist management information system (MIS),
decision support systems (DSS), expert system (ES), scenario development (SD), material

30
flow analysis (MFA), life cycle assessment (LCA) or life cycle inventory (LCI), risk
assessment (RA), environmental impact assessment (EIA), and strategic environmental
assessment (SEA), socio-economic assessment (SoEA), and sustainable assessment (SA).

The integration of these systems analysis techniques plays an important role in SWM
system because these alternatives help to tackle challenges for cost-benefit-risks trade-offs
through a wide range of comparative analysis (N.-B. Chang et al., 2015).

2.1.3 Multi criteria decision making models

MCDM was a discipline aimed at supporting decision makers and stakeholders deal with
numerous conflicting objectives and contradictory impact factors. The method was used to
reach optimal and compromise solutions and improve the quality of made decision by
satisfying multiple criteria. MCDM include multi attributes and multiple objectives, which
are known as multi attributes decision making (MADM) and multi objective decision making
(MODM) in decision science. MADM issues are based on classical decision analysis to
conduct trade-offs among a finite number of alternatives with respect to a set of attributes of
the system such as cost, risks or emission. MADM is applied as a caliber for decision makers
to choose among different alternatives, meanwhile, MODM is applied to problems with multi
conflicting objectives to identify a feasible alternative that yields the most satisfactory set of
values for the objective functions (N.-B. Chang et al., 2015). MODM is not applied to
problems with predetermined alternatives, rather than selects the feasible space of solution by
considering the various interactions between constraints and objectives, which then used to
determine a compromised and optimal solution.

Previous studies applied MODM model in waste management, N.-B. Chang et al. (1996)
developed a MODM model for reasoning the potential conflict between environmental and
economic goals and for evaluating sustainable strategies for waste management in a
metropolitan region; Costi et al. (2004) proposed a MODM model that aims at presenting the
structure and the application of a decision support system (DSS) designed to help decision
makers of a municipality in the development of incineration, disposal, treatment and
recycling integrated programs; and Harijani et al. (2017) develop a multi-objective mixed
integer linear programming model in which the economic, environmental and social
dimensions of sustainability are concurrently balanced for sustainable recycling of MSW.

31
 However, most of the MCDM models developed for SWM have serve drawbacks as
pointed out by Morrissey et al. (2004) and N.-B. Chang et al. (2011). Firstly, the models have
limitations in consideration the complete waste management cycle, from the generation of
waste through to final disposal. Most are only concerned with refining the actual multi-
criteria technique itself. In addition, most of models considered economic and environmental
aspects, but very few of them considered social aspects for a sustainable SWM which
requires all three aspects. The non-involvement of the people who generate the waste, (i.e. the
general public) in a meaningful way in the decision making process is a major shortcoming of
these models and as a result, it is contended that none of the models can be considered to be
fully sustainable. Another identified weakness of the current models, is that no model
identified considers the involvement of all relevant stakeholders, namely the government, the
local authorities, the technical experts and the community.

2.2 MSW characterization

2.2.1 Waste composition analysis

Data describing the actual waste flow is basic input in discussions on any of waste
management aspects such as technical design of collection system, environmental objectives,
operating costs, types of recycling material economic incentives, information strategies, etc.
Therefore, reliable waste generation and composition data are useful (Dahlén et al., 2008).
The quality of waste composition data are highly affected by the sampling procedure and
method applied (Petersen et al., 2004). Therefore, the absence of international standards for
solid waste characterization has led to a variety of sampling and sorting approaches, making a
comparison of results between studies challenging (Dahlén et al., 2008; Edjabou et al., 2015).

There are basically two approaches for analyzing waste stream composition: The
material flow approach and the output approach. The material flow approach looks at the
production and the life cycle of the product to estimate the composition of the solid waste
stream. It more applicable to large scale study and geographical area like the entire country,
rather than local studies (McCauley-Bell et al., 1997). An advantage of the material flow is
the broad geographical scope for which the solid waste stream can be estimated. However, a
fact of this method is that the focus is on product categories, which is not waste stream
categories and it can exclude some significant waste components because its generation is out
of product sector.
32
 The output approach, on the other hand, focus on components of the waste stream and
can provide information regarding the waste composition, generation prior to separation or
disposal. Direct sampling at the source and Vehicle load sampling are the most popular
waste composition sampling method of the output approach for determining the composition
of the MSW (McCauley-Bell et al., 1997; Tchobanoglous et al., 1993). The former method
can provide the waste information related to specific waste producers or waste data to be
associated with the specific area, while the latter is more about giving the general data of
waste composition and hardly to attribute accurately to the geographical areas or types of
generation sources.

Vehicle load sampling method is often carried out by sampling the waste received at
transfer stations, landfill sites or at treatment facilities (N.-B. Chang et al., 2008; Wagland et
al., 2012). The most advantage of this method is to be enable to reduce the amount of samples
and less efforts for samples collection. Nevertheless, beside the drawback of not presenting
the generation source’s charateristics, the variation of waste composition and generation is
significant. Weight, fraction and characteristic of waste could be changed due to the
compaction mechanism on most of modern collection vehicles. For instance, waste fractions
sampled from such compact vehicles have been considerably difficult in distinguishing
individual material fractions during manual due to the effect of mechanical stress and
blending. Water lost and cross-contamination between individual fractions may occur,
leading to further inaccuracies, error or variation of the study’s results.

Table 2. 1. Method for waste composition analysis

Method description No. Categories Reference Applied

recommendation
Vehicle load sampling
The American Society for Testing and At least 13 (ASTM, (N.-B. Chang et
Materials (ASTM D5231 – 92): Standard 2003) al., 2008;
Test Method for Determination of the Gidarakos et
Composition of Unprocessed Municipal al., 2006)
Solid Waste
A methodology was modeled into a 12 – 52 with (McCauley- (Sharma et al.,
knowledge-based system to guide users and median of 28 Bell et al., 2007)

33
promote consistency in MSW composition categories 1997)
studies in the state of Florida.
The Nordic Innovation Center provided a 11 main categories (Nordtest, In Sweden
standard method for sampling and 1995) (Dahlén et al.,
characterization of municipal solid waste. 2008)
At source sampling
The Nordtest method also provide a 11 main categories (Nordtest, -
guideline for sampling waste from 1995)
household
A method for stratified sampling of 8 categories (Mbande, (Mbande, 2003)
household waste based on socio-economic 2003)
conditions
A method for selecting sampling areas that 6 categories (Burnley et (Burnley et al.,
takes account of geographical, waste al., 2007) 2007)
management and socio-demographic
factors to ensure that the individual samples
taken are representative of their local area
and the aggregated samples are
representative of an entire region.

Collecting waste directly from individual households or sources of waste generation has
many advantages, it allows for the collection of personal data from each of the participating
household or it allow the waste data to be associated with the specific source (Dahlén et al.,
2008). Some previous studies conducted collecting waste from household sampling to
compare the characteristics of waste from different types of sources such as single/multi-
family (S. Lebersorger et al., 2011; S Lebersorger et al., 2003), family incomes (Bolaane et
al., 2004), etc. In addition, waste analysis at source can reduce the difficulties in degradable
waste charaterisation analysis such as food and vegetable wastes, which may decompose
quickly, contaminating the entire waste stream and making it difficult to characterize (Dangi
et al., 2008). To ensure uniform coverage of the geographical area, stratified sampling which
divide the study area into non-overlapping sub-areas with similar characteristics is often
applied (Dangi et al., 2011; Philippe et al., 2009; Sharma et al., 2007). Table 2.1 describes
some method for solid waste composition analysis applied for vehicle load sampling and at
source sampling.

34
 The limited studies conducted at the source of generation have faced difficulties using
the correct statistical tools (Dangi et al., 2008). One of the common errors that can occur is
the situation in which researchers equate the number of physical samples collected to the
number of statistical samples, or samples required to obtain statistically meaningful results
(Bolaane et al., 2004). On the other hand, the advantage of statistical tool was ignore in the
study in Dublin, Dennison et al. (1996) reported a simple random sampling formula with
post-stratification that gave 384.2 as the required statistical sample size and yet the authors
utilized a large initial sample with 1500 total households. Moreover, the examination of
normal distribution of collected data was also ignored in some studies. To apply appropriate
statistical analysis technique, the type of data distribution is relatively important.

For number of waste categories, recommendation of various methods are varied widely.
Most of methods suggest a small number of primary categories (main compositions) and a
bigger number of secondary categories (sub-compositions) (N. P. Thanh, 2010). However,
the number of waste categories applied in a study is often depended on the purpose of the
particular study. For instance, a study for MSW management planning often use 9-12 primary
categories but a Life Cycle Analysis (LCA) or Material Flow Analysis (MFA) study, will
separate waste to more (up to 100) categories (N. Thanh et al., 2012). According to Dahlén et
al. (2008), a number of waste composition of about 10, as far as possible based on the
physical characteristics would limit the misunderstanding and reduce errors in sampling
process. Also, limited waste categories will facilitate comparisons among studies, over
temporal and spatial conditions.

2.2.2 Prognosis waste generation modelling

One of the challenges faced by local governments is the prognosis of solid waste
quantities to have appropriate actions and plan (Ghinea et al., 2016). Apparently, Predicting
waste generation is increasingly essential in waste collection planning, waste treatment
strategies and establishing waste policies toward a sustainable waste management system
(Abbasi et al., 2012; Chen et al., 2000; N. P. Thanh et al., 2011). Lack of research and
method on waste generation estimation has led to a considerable challenge in municipal
waste management in Viet Nam.

Various modelling techniques were applied to the development of solid waste generation
predictive models. According to Peter Beigl et al. (2008), multivariate methods, such as
35
system dynamics and input-output analyses, are very complex due to the numerous
interactions between parameters and the main problem of these methods is difficult to
achieve model validations. Linear regression analysis, on the other hand, was more popular in
the application of waste generation estimation (Bach et al., 2004; Peter Beigl et al., 2008;
Buenrostro et al., 2001; Ghinea et al., 2016; N. P. Thanh et al., 2010). The term linear can
give the casual observer the impression that linear models can only handle simple datasets. In
fact, linear models can easily be expanded and modified to handle complex datasets in reality.
Linear regression analysis is useful in empirical investigations and prediction rather than
studying on the theoretical relationship between variables, which are better done by nonlinear
regression analysis (Faraway, 2005).

Previous studies had modelled the total municipal waste generation using various
variables and reported that municipal waste generation rate significantly correlated with
economic factors. At regions and country level, Hockett et al. (1995) created a model of the
demographic, economic and structural determinants of per capita waste generation. The study
found that the per capita retail sales and tipping fees are the significant determinants of waste
generation, and demographic variables were not significant as correlates of waste products.
Thøgersen (1996), studied on data of 18 OECD countries, showed that GDP per capita (Gross
Domestic Products) explained 50% variations of per capita waste generation (R 2 = 0.5) based
on simple linear regression analysis. Exponential and polynomial linear regression analysis in
this research proved that there was a significant correlation between GDP per capita and
waste produced per capital, and the coefficient of determination was slightly higher than
simple linear regression. Daskalopoulos et al. (1998) provided predictive models for EU and
USA for the total amount of waste arising annually with GDP and population acted as
predictor variables in polynomial equations. Another research estimated MSW generation by
multivariate linear regression model in consideration of the national gross domestic product
per capita (GDP), the infant of mortality rate per 1000 births, the percentage of the population
aged 15 to 59 years, the average household size and the life expectancy at birth as predictor
variables (P Beigl et al., 2004).

In consideration of factors affecting to household waste generation, Dennison et al.

(1996) and Abu Qdais et al. (1997) estimated waste generation rate per capita based on
household size. Abu Qdais et al. (1997) reported that the correlation between waste

36
generation rate and the number of people in the family was weak (r=0.33), but there was a
strong relationship between the waste amount and annual rental rate (r=0.83) which attributed
to household income. As Thøgersen (1996), Abu Qadis and Dennison did not give any
models to estimate per capita waste generation rate based on related factors. S Lebersorger et
al. (2003) found that the correlation between the quantities of residual waste with the
household type and age was significant in multi-family dwellings. A significant positive
correlation was also found between the number of rooms and solid-waste generation in a
study of Monavari et al. (2012).

Multivariate linear regression analysis was applied widely in waste generation

forecasting research (Grazhdani, 2016). Also, the stepwise analysis normally is utilised to
ensure that the final regression model provides the best fit (Bdour et al., 2007; Y. F. Chang et
al., 2007; Hamby, 1994; S. Lebersorger et al., 2011; Shamshiry et al., 2014). However, the
stepwise regression analysis was recommended no to use because it may not correctly
determine the best variable set and tend to yield results that are not replicable (Derksen et al.,
1992; Hamby, 1994; Thompson, 1995). In addition, the diagnostic checking to verify the
statistical adequacy of the model was not carried out (Bdour et al., 2007; Karpušenkaitė et al.,
2016; Shamshiry et al., 2014) and model validation with a new dataset also got lack of
attention (Buenrostro et al., 2001; Ghinea et al., 2016; S. Lebersorger et al., 2011). The two
analysis above was essential to evaluate the reliability and performance of a linear regression
model.

Benítez et al. (2008) developed a mathematical model to predict residential solid waste
generation using simple linear and multivariate linear regression with continuous variables
(household income) and discrete variables (education, household size). The result of this
study showed that the model selected could explain 51% of the variation of waste generation
rate, but used all predictor variables in the analysis. Apparently in multivariate linear
regression models, the higher number of independent variables, the higher value of the
coefficient of determination. Therefore, choosing the model based on the maximum
coefficient of determination is not feasible options when dealing with a lot of independent
variables.

37
 Thus, providing a reliable model with appropriate diagnostic tests for the hypothesis of
linear assumptions and a validation method is crucial for helping the decision and policy
makers and related stakeholders in forecasting the quantity of waste from households.

2.3 MSW Optimization modeling

2.3.1 Optimization models

Optimization models are the core of the systems engineering modeling approach. Single
objective models aim to seek for the optimal solution associated with a single objective and
numerous technical, environmental, financial and managerial constraints. Normally, single
objective model helps to solve the economic issues such as the cost minimization,
maximizing total benefits and formulated by deterministic methods including linear
programming (Huang et al., 1992), non-linear programming (Costi et al., 2004; Wu et al.,
2006), mixed-integer programming (Badran et al., 2006; Das et al., 2015) or combination of
them (Fiorucci et al., 2003).

Multi-objective programming (MOP) models were developed and widely applied to deal
with the need of multiple objectives such as the minimization of total cost and maximization
of recycled waste, or minimization of pollution emission at the same time in waste
management studies. N.-B. Chang et al. (1996) developed a multi objective mixed integer
programming model for reasoning the potential conflict between environmental and
economic goals. Minciardi et al. (2008) developed a multi-objective optimization model for
waste flows management based on the model proposed by Costi et al. (2004) to minimize
cost, minimize emission, unrecycled waste and amount of waste to landfill with an interactive
method to get decision makers involved in various step of decision process. Cucchiella et al.
(2014) constructed a model to evaluate and quantify the effect of initiatives for diversion of
current waste from landfill in Italy. Economics and financial indicators were used to define
the profitability of waste facilities. The study proposed efficient waste to energy plant and
regional strategies of waste management for optimizing financial and environmental benefit
of the sector. Vadenbo et al. (2014) applied a sequential approach to estimate the minimum
heat requirement for the diverse solutions obtained in the Pareto front using mixed integer
linear programming formulation of the heat exchange problem.

38
2.3.2 Solving approaches

Variety of multi-objectives optimization model solving approaches were applied for

searching optimal solution. Some well-known and most used methods are mentioned as
following:

Weight method is a scalarization method and the most basic technique for solving multi-
objective problems. It combines multiple objectives into one single objective scalar function
by using weights for each single objectives. Mathematically, the method corresponds to
minimizing (or maximizing) the weighted sum of all objective functions over the set of
feasible decisions. The weighted sum will result in Pareto-optimal decision and to identify the
weights of objectives, this technique needs an experienced people who know the problem,
experts or a decision makers. Therefore, using weighted sum and experts, we can sometimes
generated all Pareto-optimal solution and out-comes by changing weighting coefficients.
Thus, it require a certain performance level from an optimization model, to set the weight for
the objectives (Lyeme et al., 2017). Moreover, if the model is non-convex, the full Pareto
front cannot be generated, thus limiting the advantage of the weighting method. Since, every
problem has different settings and every model behaves in a different way, weighted method
is not an appropriate method for solving the multi-objective problems. However, it is most
common multi-objective optimization method implemented in waste management researches
(Ahluwalia et al., 2006; Alumur et al., 2007; Galante et al., 2010; Zhang et al., 2005).

Goal programming method help to specify what objective outcomes we would like to
achieve. This method might work in a much more general setting than the method using of
weighting coefficients, but it is more complicated mathematically. This approach is using the
target levels or goals to form the feasible point, where a solution is considered as feasible if
the deviations from that reference point is minimal. Appearantly, a goal setting by the
decision maker is a more intuitive approach since it correlates to the behavior of the problem
under study. It is appropriate for the decision maker to capture the necessary elements of a
problems and formulate there into goals and constraints. However, finding the optimal set of
solutions can be tricky, since the distance from the goal can mislead the decision maker and
can produce decisions that are not Pareto-optimal. Hence, additional assumption are
necessary to have good results from the goal programming. For example, the target levels or
goals are pessimistic to limiting the use of goals to objective values that are not highly

39
unrealistic (Wierzbicki et al., 2000) and restricting the area of optimal solutions (Lyeme et al.,
2017).

Another approach for prioritization of the objectives is Lexicographic method, where the
decision maker make an order of objective functions to be minimized. After the first
objective is optimized, its value should be preserved and used as a constraint in the following
objective function minimization. The process is terminated as soon as one of the
programming problems results in a single solution. Not many studies use this solution
method. A study of Sudhir et al. (1996) used lexicographic method to seek for the optimal
solution of six objectives by comparing different priority order of the objectives.

2.3.3 Gaps in sustainable waste management optimization models

Various optimization models for optimal solution in waste management were developed
over the last decade. However, the environmental consideration in models still need to be
improved. Single objective models is addressing the environmental issues in cost or as
constraints. These are not sustainable for managing the environmental factors since they are
not directly focused on the minimization of waste emission. Thus, the narrow scope of
environmental factors as constraints in the current models mean that there is a need to go
further and include new environmental metrics. Also, multi-objective optimization models
are well addressed by presenting a decision variable for the selection of technologies rather
than environmental impacts. Because environmental impact cannot only be considered only
in incinerator emission (Minciardi et al., 2008) or only taking to account for GHG emission
(Govindan et al., 2014; Mavrotas et al., 2015) or CO2 emission (Wang et al., 2011) but also
should consider other technologies such as recycling, composting, AD, landfill, etc. and more
pollutants emission. Moreover, the environmental issue should not be considered in the
economic cost for sustainable SWM, it should be considered as independent entity in the
modeling formulation and taking the holistic approach.

In addition, the social factors should be improved up since they are not well presented in
previous models. Thus, the implementation of the objective function with social factors, or as
parameters of DM in sustainable multi-objective optimization models is inevitable. The
decision making process needs the involvement of waste generator (people who generated
waste) in fullfill the meaning of sustainability. This issue should go together with upgrading

40
the environmental issues in the objective function to include new methods of solving multi-
objective models.

2.4 Conclusions

One of difficulties in application of optimization model in SWM is to reflex the

complexity of real system which requires a reliable database of waste characterization and
intensive information as input data and parameters of the models. Waste generation and
characterization analysis plays an important role for collecting input data in waste
management modeling. However, within the literatures, there is still the lack of appropriate
statistical analysis and test might affect the results of waste characterization studies.
Moreover, the relevant factors of waste generation as well as the prognosis model expected to
be useful for decision makers, researchers, stakeholders, recyclers for planning and
improving the current status of waste collection and management. Nevertheless, studies is
still lagged behind in Vietnam and the models’ reliability and performance evaluation has
gotten lack of attention.

System analysis approach and MOP model is a useful tool in decision science to support
decision making in solid waste management. Most of models applied in waste management
did not considered the waste management cycle, from waste generation to final disposal,
except the LCA. Thus, the combination of LCA and LCI in optimization modeling of
sustainable waste management can help to solve the weakness in analysis of waste cycle in
OM models. In addition, it also enable the model to consider more kind of pollutants
emission from all process and treatment facilities of waste management system. However, in
this dissertation, the exact calculation of LCA or LCI is not the purpose of this work. Thus, in
order to consider the environmental aspects and waste stream in the system, available data in
the existing researches and the reliable scientific reports in LCI and LCA will be used as
references for model parameters.

Lastly, few previous studies have considered public participation in the decision-making
process. The public is only apprised or takes part in discussion, and has little effect on
decision making in most research efforts (Hung et al., 2007). Especially, the non-involvement
of the people who generate the waste, (i.e. the general public) in a meaningful way in the
decision making process is a major shortcoming of these models. Thus, a method of modeling
which can take into account the participation of all relevant stakeholders, authorities and
41
citizens is crucial for supporting decision making of a sustainable waste management system.
This doctoral dissertation will develop optimization models with an appropriate approach that
fully considering the environmental, economic and social factors to provide optimal solution
of sustainable waste management in a case study in Viet Nam.

2.5 References

Abbasi, M., Abduli, M., Omidvar, B., & Baghvand, A. (2012). Forecasting municipal solid
waste generation by hybrid support vector machine and partial least square model.
International Journal of Environmental Research, 7(1), 27-38.
Abu Qdais, H. A., Hamoda, M. F., & Newham, J. (1997). Analysis of Residential Solid
Waste At Generation Sites. Waste Management & Research, 15(4), 395-405.
doi:10.1177/0734242x9701500407
Ahluwalia, P. K., & Nema, A. K. (2006). Multi-objective reverse logistics model for
integrated computer waste management. Waste Manag Res, 24(6), 514-527.
Alumur, S., & Kara, B. Y. (2007). A new model for the hazardous waste location-routing
problem. Computers & Operations Research, 34(5), 1406-1423.
ASTM. (2003). Standard Test Method for Determination of the Composition of
Unprocessed Municipal Solid Waste ASTM D5231-92. West Conshohocken, PA: ASTM
International.
Bach, H., Mild, A., Natter, M., & Weber, A. (2004). Combining socio-demographic and
logistic factors to explain the generation and collection of waste paper. Resources,
Conservation and Recycling, 41(1), 65-73.
doi:http://dx.doi.org/10.1016/j.resconrec.2003.08.004
Badran, M. F., & El-Haggar, S. M. (2006). Optimization of municipal solid waste
management in Port Said – Egypt. Waste Management, 26(5), 534-545.
doi:http://doi.org/10.1016/j.wasman.2005.05.005
Bdour, A., Altrabsheh, B., Hadadin, N., & Al-Shareif, M. (2007). Assessment of medical
wastes management practice: a case study of the northern part of Jordan. Waste Manag, 27(6),
746-759. doi:10.1016/j.wasman.2006.03.004
Beigl, P., Lebersorger, S., & Salhofer, S. (2008). Modelling municipal solid waste
generation: A review. Waste Management, 28(1), 200-214.
doi:10.1016/j.wasman.2006.12.011
Beigl, P., Wassermann, G., Schneider, F., & Salhofer, S. (2004). Forecasting municipal
solid waste generation in major European cities. Paper presented at the iEMSs 2004
International Congress: Complexity and Integrated Resources Management, Osnabrueck,
Germany.
Benítez, S. O., Lozano-Olvera, G., Morelos, R. A., & Vega, C. A. d. (2008). Mathematical
modeling to predict residential solid waste generation. Waste Management, 28, Supplement 1,
S7-S13. doi:http://dx.doi.org/10.1016/j.wasman.2008.03.020

42
 Bolaane, B., & Ali, M. (2004). Sampling Household Waste at Source: Lessons Learnt in
Gaborone. Waste Management & Research, 22(3), 142-148. doi:10.1177/0734242x04044970
Buenrostro, O., Bocco, G., & Vence, J. (2001). Forecasting Generation of Urban Solid
Waste in Developing Countries—A Case Study in Mexico. Journal of the Air & Waste
Management Association, 51(1), 86-93. doi:10.1080/10473289.2001.10464258
Burnley, S. J., Ellis, J. C., Flowerdew, R., Poll, A. J., & Prosser, H. (2007). Assessing the
composition of municipal solid waste in Wales. Resources, Conservation and Recycling,
49(3), 264-283. doi:http://dx.doi.org/10.1016/j.resconrec.2006.03.015
Chang, N.-B., & Davila, E. (2008). Municipal solid waste characterizations and
management strategies for the Lower Rio Grande Valley, Texas. Waste Management, 28(5),
776-794. doi:http://dx.doi.org/10.1016/j.wasman.2007.04.002
Chang, N.-B., & Pires, A. (2015). Systems Engineering Tools and Methods for Solid
Waste Management Sustainable Solid Waste Management (pp. 235-299): John Wiley & Sons,
Inc.
Chang, N.-B., Pires, A., & Martinho, G. (2011). Empowering Systems Analysis for Solid
Waste Management: Challenges, Trends, and Perspectives. Critical Reviews in
Environmental Science and Technology, 41(16), 1449-1530.
doi:10.1080/10643381003608326
Chang, N.-B., & Wang, S. F. (1996). Solid Waste Management System Analysis by
Multiobjective Mixed Integer Programming Model. Journal of Environmental Management,
48(1), 17-43. doi:http://dx.doi.org/10.1006/jema.1996.0064
Chang, Y. F., Lin, C. J., Chyan, J. M., Chen, I. M., & Chang, J. E. (2007). Multiple
regression models for the lower heating value of municipal solid waste in Taiwan. Journal of
Environmental Management, 85(4), 891-899.
doi:http://dx.doi.org/10.1016/j.jenvman.2006.10.025
Chen, H. W., & Chang, N.-B. (2000). Prediction analysis of solid waste generation based
on grey fuzzy dynamic modeling. Resources, Conservation and Recycling, 29(1-2), 1-18.
doi:10.1016/s0921-3449(99)00052-x
Costi, P., Minciardi, R., Robba, M., Rovatti, M., & Sacile, R. (2004). An environmentally
sustainable decision model for urban solid waste management. Waste Management, 24(3),
277-295.
Cucchiella, F., D’Adamo, I., & Gastaldi, M. (2014). Strategic municipal solid waste
management: A quantitative model for Italian regions. Energy Conversion and Management,
77, 709-720. doi:http://dx.doi.org/10.1016/j.enconman.2013.10.024
Dahlén, L., & Lagerkvist, A. (2008). Methods for household waste composition studies.
Waste Management, 28(7), 1100-1112. doi:http://dx.doi.org/10.1016/j.wasman.2007.08.014
Dangi, M. B., Pretz, C. R., Urynowicz, M. A., Gerow, K. G., & Reddy, J. M. (2011).
Municipal solid waste generation in Kathmandu, Nepal. Journal of Environmental
Management, 92(1), 240-249. doi:http://dx.doi.org/10.1016/j.jenvman.2010.09.005
Dangi, M. B., Urynowicz, M. A., Gerow, K. G., & Thapa, R. B. (2008). Use of stratified
cluster sampling for efficient estimation of solid waste generation at household level. Waste
Management & Research, 26(6), 493-499. doi:10.1177/0734242x07085755

43
 Das, S., & Bhattacharyya, B. K. (2015). Optimization of municipal solid waste collection
and transportation routes. Waste Management, 43, 9-18.
doi:http://doi.org/10.1016/j.wasman.2015.06.033
Daskalopoulos, E., Badr, O., & Probert, S. D. (1998). Municipal solid waste: a prediction
methodology for the generation rate and composition in the European Union countries and
the United States of America. Resources, Conservation and Recycling, 24(2), 155-166.
doi:http://dx.doi.org/10.1016/S0921-3449(98)00032-9
Dennison, G. J., Dodd, V. A., & Whelan, B. (1996). A socio-economic based survey of
household waste characteristics in the city of Dublin, Ireland — II. Waste quantities.
Resources, Conservation and Recycling, 17(3), 245-257. doi:http://dx.doi.org/10.1016/0921-
3449(96)01155-X
Derksen, S., & Keselman, H. J. (1992). Backward, forward and stepwise automated subset
selection algorithms: Frequency of obtaining authentic and noise variables. British Journal of
Mathematical and Statistical Psychology, 45(2), 265-282. doi:10.1111/j.2044-
8317.1992.tb00992.x
Edjabou, M. E., Jensen, M. B., Götze, R., Pivnenko, K., Petersen, C., Scheutz, C., &
Astrup, T. F. (2015). Municipal solid waste composition: Sampling methodology, statistical
analyses, and case study evaluation. Waste Management, 36, 12-23.
doi:http://dx.doi.org/10.1016/j.wasman.2014.11.009
F.R. McDougall, P. R. W., & Hindle, M. F. a. P. (2001). Chapter 2: Integrated Waste
Management Integrated Solid Waste Management: A life-cycle Inventory (second ed., pp.
15-31). MPG Books Ltd, Bodmin, Cornwall: Blackwell Publishing Company.
Faraway, J. J. (2005). Linear models with R. Boca Raton, FL. [etc.]: Chapman &
Hall/CRC.
Fiorucci, P., Minciardi, R., Robba, M., & Sacile, R. (2003). Solid waste management in
urban areas: Development and application of a decision support system. Resources,
Conservation and Recycling, 37(4), 301-328. doi:http://dx.doi.org/10.1016/S0921-
3449(02)00076-9
Galante, G., Aiello, G., Enea, M., & Panascia, E. (2010). A multi-objective approach to
solid waste management. Waste Management, 30(8–9), 1720-1728.
doi:http://dx.doi.org/10.1016/j.wasman.2010.01.039
Ghinea, C., Drăgoi, E. N., Comăniţă, E.-D., Gavrilescu, M., Câmpean, T., Curteanu, S., &
Gavrilescu, M. (2016). Forecasting municipal solid waste generation using prognostic tools
and regression analysis. Journal of Environmental Management, 182, 80-93.
doi:http://dx.doi.org/10.1016/j.jenvman.2016.07.026
Gidarakos, E., Havas, G., & Ntzamilis, P. (2006). Municipal solid waste composition
determination supporting the integrated solid waste management system in the island of Crete.
Waste Management, 26(6), 668-679. doi:http://dx.doi.org/10.1016/j.wasman.2005.07.018
Govindan, K., Jafarian, A., Khodaverdi, R., & Devika, K. (2014). Two-echelon multiple-
vehicle location–routing problem with time windows for optimization of sustainable supply
chain network of perishable food. International Journal of Production Economics, 152, 9-28.
doi:http://dx.doi.org/10.1016/j.ijpe.2013.12.028

44
 Grazhdani, D. (2016). Assessing the variables affecting on the rate of solid waste
generation and recycling: An empirical analysis in Prespa Park. Waste Management, 48, 3-13.
doi:http://dx.doi.org/10.1016/j.wasman.2015.09.028
Hamby, D. M. (1994). A review of techniques for parameter sensitivity analysis of
environmental models. Environmental Monitoring and Assessment, 32(2), 135-154.
doi:10.1007/bf00547132
Harijani, A. M., Mansour, S., & Karimi, B. (2017). A multi-objective model for
sustainable recycling of municipal solid waste. Waste Management & Research, 35(4), 387-
399. doi:doi:10.1177/0734242X17693685
Hockett, D., Lober, D. J., & Pilgrim, K. (1995). Determinants of Per Capita Municipal
Solid Waste Generation in the Southeastern United States. Journal of Environmental
Management, 45(3), 205-217. doi:http://dx.doi.org/10.1006/jema.1995.0069
Huang, G., Baetz, B. W., & Patry, G. G. (1992). A grey linear programming approach for
municipal solid waste management planning under uncertainty. Civil Engineering Systems,
9(4), 319-335.
Hung, M.-L., Ma, H.-w., & Yang, W.-F. (2007). A novel sustainable decision making
model for municipal solid waste management. Waste Management, 27(2), 209-219.
doi:http://dx.doi.org/10.1016/j.wasman.2006.01.008
Jeffrey K, S. (2010). Sustainable waste management systems. Journal of Cleaner
Production, 18(16–17), 1639-1651.
Karpušenkaitė, A., Denafas, G., & Ruzgas, T. (2016). Forecasting Hazardous Waste
Generation Using Short Data Sets: Case Study of Lithuania. Science–Future of
Lithuania/Mokslas–Lietuvos Ateitis, 8(4), 357-364.
Lebersorger, S., & Beigl, P. (2011). Municipal solid waste generation in municipalities:
quantifying impacts of household structure, commercial waste and domestic fuel. Waste
Manag, 31(9-10), 1907-1915. doi:10.1016/j.wasman.2011.05.016
Lebersorger, S., Schneider, F., & Hauer, W. (2003). Waste generation in households–
models in theory and practical experience from a case study of multifamily dwellings in
Vienna. Proceedings in Sardinia.
Lyeme, H. A., Mushi, A., & Nkansah-Gyekye, Y. (2017). Review of multi-objective
optimization models for solid waste management systems with environmental considerations.
Journal of Mathematical and Computational Science, 7(1), 150.
Marshall, R. E., & Farahbakhsh, K. Systems approaches to integrated solid waste
management in developing countries. Waste Management(0).
doi:10.1016/j.wasman.2012.12.023
Mavrotas, G., Gakis, N., Skoulaxinou, S., Katsouros, V., & Georgopoulou, E. (2015).
Municipal solid waste management and energy production: Consideration of external cost
through multi-objective optimization and its effect on waste-to-energy solutions. Renewable
and Sustainable Energy Reviews, 51, 1205-1222.
doi:http://dx.doi.org/10.1016/j.rser.2015.07.029
Mbande, C. (2003). Appropriate approach in measuring waste generation, composition and
density in developing areas : technical paper45(3), p.2-10. Retrieved from

45
http://reference.sabinet.co.za/webx/access/electronic_journals/civileng/civileng_v45_n3_a1.p
df
McCauley-Bell, P., Reinhart, D. R., Sfeir, H., & Ryan, B. O. T. (1997). Municipal Solid
Waste Composition Studies. Practice Periodical of Hazardous, Toxic, and Radioactive Waste
Management, 1(4), 158-163. doi:doi:10.1061/(ASCE)1090-025X(1997)1:4(158)
Minciardi, R., Paolucci, M., Robba, M., & Sacile, R. (2008). Multi-objective optimization
of solid waste flows: Environmentally sustainable strategies for municipalities. Waste
Management, 28(11), 2202-2212.
Monavari, S., Omrani, G., Karbassi, A., & Raof, F. (2012). The effects of socioeconomic
parameters on household solid-waste generation and composition in developing countries (a
case study: Ahvaz, Iran). Environmental Monitoring and Assessment, 184(4), 1841-1846.
doi:10.1007/s10661-011-2082-y
Morrissey, A. J., & Browne, J. (2004). Waste management models and their application to
sustainable waste management. Waste Management, 24(3), 297-308.
Nordtest. (1995). Solid waste, municipal: Sampling and characterisation (NT ENVIR 001).
Nordtest method (Publication no.
http://www.nordtest.info/index.php/methods/environment/item/solid-waste-municipal-
sampling-and-characterisation-nt-envir-001.html?category_id=24). http://www.nordtest.info
Petersen, L., Dahl, C. K., & Esbensen, K. H. (2004). Representative mass reduction in
sampling—a critical survey of techniques and hardware. Chemometrics and Intelligent
Laboratory Systems, 74(1), 95-114. doi:http://dx.doi.org/10.1016/j.chemolab.2004.03.020
Philippe, F., & Culot, M. (2009). Household solid waste generation and characteristics in
Cape Haitian city, Republic of Haiti. Resources, Conservation and Recycling, 54(2), 73-78.
doi:http://dx.doi.org/10.1016/j.resconrec.2009.06.009
Shamshiry, E., Bin Mokhtar, M., & Abdulai, A. (2014). Comparison of Artificial Neural
Network (ANN) and Multiple Regression Analysis for Predicting the Amount of Solid Waste
Generation in a Tourist and Tropical Area—Langkawi Island. Paper presented at the
proceeding of International Conference on Biological, Civil, Environmental Engineering
(BCEE).
Sharma, M., & McBean, E. (2007). A methodology for solid waste characterization based
on diminishing marginal returns. Waste Manag, 27(3), 337-344.
doi:10.1016/j.wasman.2006.02.007
Sudhir, V., Muraleedharan, V., & Srinivasan, G. (1996). Integrated solid waste
management in urban India: a critical operational research framework. Socio-Economic
Planning Sciences, 30(3), 163-181.
Tchobanoglous, G., Theisen, H., & Vigil, S. (1993). Integrated solid waste management:
engineering principles and management issues: McGraw-Hill, Inc.
Thøgersen, J. (1996). Theme: Towards Green Business SystemsWasteful food
consumption: Trends in food and packaging waste. Scandinavian Journal of Management,
12(3), 291-304. doi:http://dx.doi.org/10.1016/0956-5221(96)00011-5

46
 Thanh, N., & Matsui, Y. (2012). An evaluation of alternative household solid waste
treatment practices using life cycle inventory assessment mode. Environmental Monitoring
and Assessment, 184(6), 3515-3527. doi:10.1007/s10661-011-2205-5
Thanh, N. P. (2010). A Study on Eveluation Methodologies for Household Solid Waste
Management toward a Sustainable Society in Vietnam. (Doctor), Okayama University,
Okayama, Japan.
Thanh, N. P., & Matsui, Y. (2011). Municipal solid waste management in Vietnam: Status
and the strategic actions. International Journal of Environmental Research, 5(2), 285-296.
Thanh, N. P., Matsui, Y., & Fujiwara, T. (2010). Household solid waste generation and
characteristic in a Mekong Delta city, Vietnam. Journal of Environmental Management,
91(11), 2307-2321. doi:http://dx.doi.org/10.1016/j.jenvman.2010.06.016
Thompson, B. (1995). Stepwise Regression and Stepwise Discriminant Analysis Need Not
Apply. Paper presented at the The Annual Meeting of the American Educational Research
Association (San Francisco, CA, April 18-22, 1995), San Francisco, CA.
http://files.eric.ed.gov/fulltext/ED382635.pdf
Vadenbo, C., Hellweg, S., & Guillén-Gosálbez, G. (2014). Multi-objective optimization of
waste and resource management in industrial networks – Part I: Model description. Resources,
Conservation and Recycling, 89, 52-63.
doi:http://dx.doi.org/10.1016/j.resconrec.2014.05.010
Wagland, S. T., Veltre, F., & Longhurst, P. J. (2012). Development of an image-based
analysis method to determine the physical composition of a mixed waste material. Waste
Management, 32(2), 245-248. doi:http://dx.doi.org/10.1016/j.wasman.2011.09.019
Wang, F., Lai, X., & Shi, N. (2011). A multi-objective optimization for green supply chain
network design. Decision Support Systems, 51(2), 262-269.
doi:http://dx.doi.org/10.1016/j.dss.2010.11.020
Wierzbicki, A., Makowski, M., & Wessels, J. (2000). Model-based decision support
methodology with environmental applications: Kluwer Academic Dordrecht, The
Netherlands.
Wu, X. Y., Huang, G. H., Liu, L., & Li, J. B. (2006). An interval nonlinear program for the
planning of waste management systems with economies-of-scale effects—A case study for
the region of Hamilton, Ontario, Canada. European Journal of Operational Research, 171(2),
349-372. doi:http://dx.doi.org/10.1016/j.ejor.2005.01.038
Zarate, M. A., Slotnick, J., & Ramos, M. (2008). Capacity building in rural Guatemala by
implementing a solid waste management program. Waste Manag, 28(12), 2542-2551.
doi:10.1016/j.wasman.2007.10.016
Zhang, M., Ma, Y., & Weng, K. (2005). Location-routing model of hazardous materials
distribution system based on risk bottleneck. Paper presented at the Services Systems and
Services Management, 2005. Proceedings of ICSSSM'05. 2005 International Conference on.

47
 3. METHODOLOGY AND PROPOSED CASE STUDY
3.1 Methodology
3.1.1 Waste generation and composition study
3.1.1.1 Stratified random sampling method
The purpose of sample survey design is to maximize the amount of information for a given
cost. Simple random sampling, the core sampling design, often provides useful estimates of
population quantities at low cost. According to Scheaffer et al. (2011) stratified random sampling,
which in many instances increases the amount of information for a given cost.

“A stratified random sample is one obtained by separating the population elements into
nonoverlapping groups, called strata, and then selecting a simple random sample from each
stratum” (Scheaffer et al., 2011).

The procedure of stratified random sampling method is presented as follows:

The first step in the selection of a stratified random sample is to specify the strata clearly,
then each sampling unit of the population is placed into its appropriate stratum. For example, we
suppose to stratify the sampling for household waste which is different between rural and urban
areas. Hence, to specify what is meant by urban and rural is essential so that each sampling unit
falls into only one stratum.

After the sampling areas have been divided into strata, we select a simple random sample
from each stratum and the statistical sample size will be calculated based on the guideline in
(Nordtest, 1995). We must be certain that the samples selected from the strata are independent.
That is, different random sampling schemes should be used within each stratum so that the
observations chosen in one stratum do not depend on those selected in another.

3.1.1.2 Waste sample collection

To collect waste samples for waste generation and composition analysis, we applied door-to-
door waste collecting method from every single household which was chosen as survey sample.
The detailed procedure will be explained in section 4.

3.1.1.3 Classification categories of solid waste

Most methods suggest a limited number of primary categories (also called principal
components), and a large number of secondary, tertiary, etc. categories (subcomponents), which
48
are more or less applied depending on the purpose of a particular study. In addition, the statistical
significance will decrease when the number of components increases, which is described above
in the discussion about the number of sub-samples needed. The aim of this study is to provide
detailed information on waste composition to support the development of sustainable waste
management plan. Thus, in this study author will choose the number of main categories is 10 and
the number of secondary categories is 18 according to the guideline in Nordtest (1995), based on
physical appearance characteristics and usage purpose. The detailed information of categories
will be described in the next chapter.

For component analyses, author chose the mother sample (the lot) from each stratum to
cover consecutively 14 days, and the size of each analysed sample is more than 90 kg (Sfeir et al.,
1999).

3.1.1.4 Questionnaire survey

A questionnaire survey was carried out with the face-to-face interview at the house of
families involved to waste sampling program to obtain data reflecting to socioeconomic factors
and demographic information of the household. This information will be used to evaluate the
correlation of household waste generation with related factors and modelling prognosis waste
generation model.

3.1.1.5 Laboratory analysis

In this study, physical characteristics of waste including specific weight, the moisture
content will be analysed by ASTM-E1109 (2009) and ASTM-E949 (2004), respectively. The
chemical elements and features such as volatile matter and ash content and energy content are
analysed by (ASTM-E897-88, 2004) and (E711-87, 2004) standard. We used CHN analyser for
analyzing the C, H and N content in waste samples. For each category such as Food, Garden,
Plastic, Cardboard, Paper, Combustible and Hazardous waste several samples were collected
randomly for analysis.

3.1.2 Mathematical modelling of waste generation

The data obtained in the previous part of this study were interpreted to independent and
dependent variables to modelling the waste generation model. A Bayesian Model Averaging is
applied to identify the significant predictor variables for estimating waste generation. This

49
solution provides optimal predictive ability by involving averaging over all possible models
(Madigan et al., 1994). Quantities of interest and parameter estimates are provided via direct
application of the principles described as follows:

The posterior distribution given data Z of the quantity of interest  , such as a model parameter
or a future observable, is defined by Eq. 3.1.

K
p(  | Z )   pr (  | M k , Z ) pr ( M k | Z )
k 1 (3.1)

Where, M1, M2, ..., Mk are the models under consideration. Eqs. 3.2 and 3.3 give the posterior
probability for model Mk and the integrated likelihood of Mk respectively.

pr ( Z | M k ) pr ( M k ) (3.2)
pr ( M k | Z )  K

 pr( Z | M
k 1
k ) pr ( M k )

pr ( Z | M k )   pr ( Z |  k , M k ) pr ( k | M k )d k (3.3)

Where,  k is the vector of parameters of model Mk, pr( k | M k ) is the prior density of the

parameters under the model, pr ( D |  k , M k ) is the likelihood, and pr ( M k ) is the prior probability
that Mk is the actual model. The BMA estimates a parameter  using Eq. 3.4.
^ K
 BMA    k pr ( M k | Z ) (3.4)
k 1

The Bayesian information criterion (BIC) approximation is formally defined in Eq. 3.5. The
BIC is used as a criterion for model selection from the set of models, and the model with the
lowest BIC approximation is preferred.

BIC = -2. log (RSSp) + p.log n (3.5)

Where,

BIC: is Bayesian information criterion

50
 RSSp: is square sum of residuals in fitting sample data to the model with p independent
variables

p: is the number of regressors including the intercept

n: is the number of observation or equivalently, the sample size

Then, Multivariate linear regression analysis will be applied to develop the prognosis model
of waste generation. The multivariate linear regression model is described as equation followed:

Y = α + β1X1 +β2X2 +β3X3 +… + ε (3.6)

Where α is the intercept term which indicates the mean of the response variable Y when all
predictor variable X is equal to 0; β, a vector of β i, is the slope of the model, and it explains the
average change in the dependent variable. The residual ε represents the difference between
estimated values and observed values. ε may include measurement error although it is often due
to the effect of unincluded or unmeasured variables (Faraway, 2005; Longnecker et al., 2001).

3.1.3 Single optimization model for evaluation of alternative waste management initiatives
3.1.3.1 Single optimization model
The most basic technique for solving OMs is a set of “mathematical programming”, a
representative technique in the field of operational research. Each mathematical programming is
based on a different mathematical structure that can only be applied to optimize a specific
problem with unique properties. Solving these models should rely on well-defined algorithms
designed to obtain the optimal solution in the problem domain. Four useful and popular methods
in mathematical programming are Linear Programming (LP), Integer Programming (IP), Non-
Linear Programming (NLP) and Dynamic Programming (DP).

The purpose of this study is proposed a model to describe the waste flow and cycle from the
generation source to final disposal with the objective of minimizing total cost. Thus, the NLP is
applied because there are nonlinear functions in the model for describing the waste flow. In
general, a nonlinear OM can be written as follows:

Min (Max): Z = C.X

Subject to: A.X ≥,=,≤ B

51
 X≥0

Where,

X: is a n×1 column vector of decision variables including n elements X 1, X2,…. Xn

A: is a m×n matrix of constants a11, a12,…, a1n,…, amn, which can be negative or positive

B: is a m×1 column vector of nonnegative constraints

C: is a 1×n row vector of constants c1, c2, …,cn, which can be negative or positive

The first equation is called objective function and the second and the third equation are
called constraints. When the objective function and/or the constraints are a nonlinear function,
the mathematical programming model is defined as a nonlinear programming model.

3.1.3.2 Analysis of waste management strategies alternatives

To analyse the feasibility of alternative waste management strategies toward to economic
concern, GHG emission reduction as well as pollution emission reduction. Classification
scenarios categorized by waste-to-landfill reduction target and waste separation efficiency
alternatives toward to minimizing total cost of the system. Thus, 12 single optimization model
based on 12 scenarios chosen were solved for comparing total cost, total pollutant emission and
GHG emission. The detailed of scenarios description is mentioned in section 6.

3.1.4 Multi-objective optimization model for MODM

3.1.4.1 . Questionnaire survey
Another simple questionnaire survey of local experts who play as authorities, waste
managers, stakeholders, environmental scientists, citizen representatives is also conducted to
obtain the idea of potential treatment technologies suitable for the socio-economic conditions of
the city as well as the important objectives and requirements of different decision makers. The
results of this survey will contribute to optimization model modelling. In this study, 20 local
experts were interviewed for the analysis.

3.1.4.2 . Multi-objective optimization model

The use of a single objective optimization approach has been recently recognized to limit the
applicability of mathematical programming models, and many multi-objective situations exist in
real-world systems. Thus, multi-objective is developed to fulfil the complicated problem in
52
reality. The complexity in MODM, some objectives permit a precise performance measurement,
such as economics, but others do not, such as aesthetics and convenience. To solve these
problems, information gathering, identification of achievable objectives, and the model building
should be addressed as a whole to bridge the gaps between real world problems and
mathematical programming models. The challenge is to adequately identify the actual objectives
by choosing not only the appropriate criteria but also the fair number of targets.

A MODM with k objective Z(x) and n decision variables x j (j∈[1;n]) and m constraints
gi(x) can be defined mathematically in a vector notation as

Maximize (Minimize) Z(x) = [Z1(x), Z2(x), Z3(x),…, Zk(x)],

Subjects to: g(x) ≤ 0

x ≥0.

However, determining the set of alternatives may require considerable effort, MODM
problems normally reflect the following needs to identify the best policy for the public (Chang et
al., 2015):

Multiple objectives with or without priority: much efforts have been made for the last
decade to develop the model with objectives other than only economic considerations.

Multiple decision makers in a participatory process: The majority of the optimization

methods presented in earlier times deal with a single decision maker, but real-world decision
making frequently involves multiple decision makers, such as a variety of stakeholders.
Behavioral and analytical models need to be developed that reflect different thinking and
preferences at the various stages of decision making associated with MODM.

When more than one objective function is used in a system engineering problem, they often
conflict. An increase in one may cause a decrease in the other and vice versa. Therefore, the
optimum solution no longer remains unique, and simultaneously maximizing together all the
conflict objectives is impossible. Optimality is thus replaced by the satisfying solution or
compromise solution. Methodologies need to systematically guide the selection of a
“compromise” or “satisficing” solution rather than a “best” solution.

53
 The concept of Pareto optimality was first introduced within the framework of welfare
economics by the Italian economist Vilfredo Prato. The decision is Pareto-optimal if we cannot
improve one of the criteria outcomes without deteriorating other criteria outcomes. In more
general cases, when we minimize some criteria, maximize others, or even do different things
with other criteria (e.g., keep them close to prescribed values), we speak about efficient decisions
and outcomes. Thus, the concept of efficiency just generalizes the concept of Pareto-optimality.
Pareto-optimality is an essentially weaker concept than single criterion optimality. Pareto-
optimality does not tell us which decisions to choose; it tells us only which decisions to avoid (A.
Wierzbicki et al., 2000).

3.1.4.3 Reference Point method for decision making

The Reference Point Method and its interactive technique were firstly proposed by A. P.
Wierzbicki (1982) and later successfully applied in engineering design, solving environmental
management problems as well as other fields of decision support (A. Wierzbicki et al., 2000).
Reference point methodology could be considered as a generalization of goal programming,
aiming at using arbitrary (not only unrealistic) goals or reference objectives and obtaining only
efficient outcomes, at the cost of avoiding norm minimization and replacing it by optimization of
a more complicated function.

The main advantages of goal programming are related to the psychologically appealing idea
that we can set a goal in objective space and try to come close to it. The basic disadvantage refers
to the fact that this idea is mathematically inconsistent with the concept of Pareto-optimality or
efficiency.

The RPM has been known as the most interactive analysis method of multiple criteria
optimisation problems. The basic concept of the interactive scheme in RPM is as follows. The
DM define their requirements regarding aspiration levels by giving a target value of each
objective outcome. Based on the reference levels provided by the DM, a special scalarizing
achievement function will be constructed, which tends to be maximised to generate an efficient
solution to the multiple criteria problem. The DM will evaluate the computed efficient solution
and compare with their aspiration level as well as the previous solution; the process allows the
modification of reference points of the DM to have a new efficient solution if he/she does not

54
satisfy with the current ones. The advantage of The RPM over other interactive methods such as
goal programming, weighted sum method is that the RPM always corresponds to a non-
dominated point of the multi-objective problem and create a chance for many DMs can come to a
good “compromise solution”. Meanwhile, the goal programming method can generate non-
efficient solutions to the multiple objective problems. On the other hand, the weighted sum
method and lexicographical method both are not proper procedures for reaching compromise
solutions as the weighted sum is fully compensatory operator that does not encode the idea of
balanced solutions (Perny et al., 2010).

3.1.5 Statistical analysis and computer software

The author calculated key statistics of waste generation and composition and also used a
variety of statistical analysis such as ANOVA analysis, linear regression, etc. The statistical
software was applied is R software (ver 3.2.0) for statistical analysis and modelling waste
generation, Microsoft Excel was used for manipulating input data.

The author also constructed and solved the optimization model by GAMS (General
Algebraic Modeling System) software (version Win 64 24.7.4).

3.2 Case study in Hoi An, Viet Nam

3.2.1 General Introduction
Hoi An is a city of Viet Nam, on the coast of the East Sea in the South Central Coast region
of Vietnam. It is located in Quang Nam Province, and it is recognized as a World Heritage
Site by UNESCO. Hoi An city has a natural land area of 6,171.25 ha with the total population of
the city in 2013 is around 93 thousand and the number population density of 1,508 people/km 2.
Hoi An city comprises 13 administrative wards including an island ward is Tan Hiep island).
Therefore, Hoi An city is a unique urban in Vietnam which including many types of urban
including urban, suburban and rural places especially it is a famous touristic city (Hoi An ancient
town), that can lead to great differences in waste generation from different locations. Also, there
are around 1,5 million of tourists coming annually to Hoi An, so the amount of commercial and
tourist waste generated is enormous and increase gradually.

55
 (source: Hoi An People committee)
Figure 3.1. The location and general map of Hoi An city
The number of tourists visiting Hoi An city (HAC) has increased over the last decade, and
its economy has developed tremendously. Rapid urbanisation and tourism development in Hoi
An have led to a considerable increase in municipal solid waste. Like many other cities in
Vietnam, the social, economic and environmental impacts of this rising waste generation are
acquiring significant attention from the society. In general, MSW in Hoi An is currently disposed
of at an open dumping landfills without taking any precautions or any operational control.
Besides, one composting plant with a capacity of 55 tonnes per day has been operating
inefficiently, and the product was not well sold in the market. The opened dump landfill and the
composting plant have caused huge adverse effects on the environment and public health.

Tourism is the most dynamic industry and the biggest economic contributor of the city. In
2013, the total income of tourist service and retail sales (from restaurants only) was 1.500 and
1.100 billion VND respectively. Meanwhile, the total industrial production value of the city was
only 212 billion VND (HASD, 2013). Also, the MSW generation increases associated with the
growth of the number of tourists (Figure 1). Thus, tourist industry might be the greatest
contributor of municipal waste generation in HAC, and it has affected strongly by the variation
of waste generation and composition. Hence, MSW management is one of the most important

56
issues of Hoi An and developing a sustainable municipal solid waste management system has
become a necessity in the City.

There was a lack of previous studies on integrated municipal waste management of the city.
Chu (2014) investigated on the co-operative model of waste management based on the
cooperation of local residences, authorities, waste managers and stakeholders in two
communities, Cam Ha and Cam Pho. The result of the study indicated that the awareness of
residence and the efficiency of waste separation at source program had been increasing. However,
the study also showed that 30% of communities had been affected by environmental pollution
related to solid waste treatment and recycling activities; 13/44 communities reported that
residences often complain about the collection system.

3.2.2 Current status of MSW management

3.2.2.1 Waste generation

Figure 3.2. MSW generation in Hoi An

MSW generation in Hoi An has increased rapidly in the past few years from 24 thousand
tonnes in 2011 to 29 thousand tonnes in 2015 (HAPWS, 2014, 2015). Besides, the tourist
industry is considered one of the most dynamic industries and the greatest contributor to the
economic development of the city (Hoang, Fujiwara, et al., 2016b). Tourist development has led
to significant influences to the environment with respects to resource consumption as well as
pollution generation including solid waste generation (Denafas et al., 2014; Espinosa Lloréns et
al., 2008; Shamshiry et al., 2011; Teh et al., 2007).

57
 In Hoi An city, one of the major impacts of tourism is the changing in generation rate and
composition of MSW. Figure 3.2 presented the amount of municipal waste generation associated
with population and tourism development of the city from 2011 to 2015. There was a sharp
increase in MSW generation in the last two years from 26 to 29 thousand tonnes, accounted for
approximately 13% of rising rate. This might be due to the steep increase in the number of
tourists came to the city from 1.75 million in 2014 to 2.25 million tourists in 2015.

Hoi A public work service Co. Ltd (HAPWSC) recently has been carrying out a waste
monitoring program to investigate the waste composition of the city. The program used sampling
from a vehicles method to analyse waste of collection trucks from each ward. According to
HAPWSC, in 2014 the average daily waste generated is 0.481 kg per capita, the total average
amount of waste generated is about 43,4 tons daily (Figure 3.3).

Minh An and Tan An ward are the two wards have the highest daily waste generation per
capita at above 0.7 kg per capita. On the other hand, Cam Thanh and Cam Kim was the lowest
source at around 0.1 kg per capita daily. Other wards are generating about 0.5 kg per capita daily.

Figure 3.3. Daily waste generation rate in Hoi An city (recorded in 2014)
3.2.2.2 Waste collection and transportation
MSW collection in Hoi An city is operated by Hoi An public work service JSC. MSW
collection in Hoi An is similar to the most of the cities in Vietnam, which is carried out as a two-

58
tier system: primary and secondary collection. In the primary collection, waste is stored and
transported from home to drop-off points, and the secondary collection involving storage and
transporting waste from collection points to treatment and disposal facilities.

Kerbside collection has been used in Hoi An MSW collection system based on the location.
The centre neighbourhoods are collected daily (Minh An, Cam Pho and Son Phong) or 6 days
per week (Tan An). Meanwhile, in the suburban and rural areas, MSW was collected three times
(Cua Dai, Cam An, Cam Nam, Thanh Ha, Cam Ha, Cam Chau, two times (Cam Thanh) and one
time a week (Cam Kim) (HAPWS, 2015).

In the city centre, residences are reminded by the collection truck with music and voice to
explain what kind of waste will be collected that day. This type of operation is not very popular
in Vietnam. Using handcart to gather waste from houses is more popular in Vietnam, and it is
applied in other areas of Hoi An city. The handcarts transport waste from household to collection
points and waste will be transported to the disposal site by trucks.

Figure 3.4. Kerbside collection in the centre of Hoi An

59
 Figure 3.5. Current MSW management in Hoi An

Table 3.1. Waste collection schedule for Hoi An city

Wards Frequency of Collection Notes
Minh An 7 days per week
Cam Pho 4 days for degradable
Son Phong 3 days for non-degradable
Cam Chau
Tan An 6 days per week and 3 days
each
Cua Dai 3 days a week
Cam An 2 days for degradable
Cam Nam 1 day for nondegradable
Thanh Ha
Cam Ha
Cam Chau
Cam Twice a week, 1 for each
Thanh
Cam Kim Once week, only non-degradable Degradable waste treated at home by
household

60
 3.2.2.3 Waste separation and recycling:
MSW in the city is composed of two groups of materials: Degradable (kitchen waste, garden
waste, etc), and nondegradable (paper, plastic, metal, dust, textile, leather, etc) waste. At present,
Hoi An is the first city in Vietnam successfully carried out waste separation at source.
Degradable and nondegradable waste is separated and collected on different days in a week. A
major difference in waste recycling activities between Vietnam and developed countries are
collecting the recyclable materials.

In developed countries, such as Japan, recyclable waste is collected and transported

separately by collecting company like other types of waste. On the other hand, in Hoi An,
recyclables were not collected officially by the waste collectors but separated and collected by
informal sectors such as scavengers and scrap collectors. Recyclable materials might be collected
at the source via buying at the households or picking up at drop-off point by scavengers. Besides,
there are many waste pickers, and scavengers work at the waste treatment plant to collect
recyclable materials; that will be sold to a bigger buyer or recyclable centre afterwards, where
waste will be separated and sold to factories as manufactured input materials.

There is few in the way of recycling initiatives of Hoi An waste managers and authorities;
thus, recycling activities are carried out by “informal” sectors such as private companies,
handicraft villages, and scavengers in all stages. The informal sectors have a significant
contribution in recycling activities and provide a livelihood to many immigrants and
marginalised people. Also, the free collection by scavengers is also expected to provide excellent
segregation of MSW as they collect the recyclable materials from the streets, bins and disposal
sites for their livelihood as well as avoid environmental costs and reduce capacity problems at
dumping and landfill sites. In Hoi An, there are currently 17 private companies household buying
of approximately 3.2 tonnes recyclable materials from scavengers per day (Chu, 2014).

However, Recycling technologies used by private companies and handicraft villages are still
traditional and backwards, thus causing serious environmental pollution problems. Especially,
some private businesses recycling waste is located inside the urban areas and thus potentially
cause serious health impact problems.

61
 3.2.2.4 Waste treatments and disposals
In Vietnam, MSW is disposed of in landfills and 85% of which either without or with
insufficient engineering standards. More than a half of MSW in Hoi An (approximately 40
tonnes per day) has dumped to an open-dumped landfill (Cam Ha landfill) in 2015. The landfill
is not engineered and no way to meet required national standards with liners, drainage system,
landfill gas control systems, leachate management system, and environmental monitoring system.
Besides the official disposal sites, the city is also being suffered from the illegal disposal of
waste in rivers, lakes, oceans, drainage channels, empty lots and roadsides.

Although there are many environmental pollution and operation problems associated with
the disposal of MSW at open dumping sites, however, it appears that landfill might continue to
be the most widely adopted practice in Hoi An in the coming few years. The reason for this
problem is caused by ineffective operation of other treatment facilities such as a composting
plant and an incinerator.

The composting plant with designed capacity of 55 tonnes per day has started operating
since 2012. According to a report of Hoi An public work service Jsc The actual amount of
degradable waste treated was average 35 tonnes with an average 4.1 tonnes (approximately 1500
tonnes per year) compost product per day (HAPWS, 2014). However, only 30% of fertiliser
could be sold and given to forestry and agriculture. As a result., there was an accumulative
amount of approximately 1.500 tonnes of fertiliser has been in stock in 2014.

Being known as a useful and cost-effective means for treatment of MSW in Vietnam but, the
product value of MSW composting in Hoi An is not high at present. Although MSW is separated
at source, there are reasons of unsuccessful composting plant in Hoi An such as inappropriate
pre-treatment technology, inadequate monitoring of the quality and quantity of the compost
being produced, and lack of attention to the market demand and to consumer feedback (D. L.
Nguyen et al., 2013; T. K. T. Nguyen, 2014). Moreover, the mix of food waste and garden waste
in waste separation program also might be a reduce the quality of compost input materials due to
the presence of dust, sands, twigs and inert materials, which contribute a major portion in garden
waste in Hoi An.

62
 The incinerator invested by the city committee has stopped operating since completely
constructed and failed the operational test in 2015. The problem was not identified; it might be a
result of the low quality of local technology or inappropriate study on MSW characteristics as
well as cost-benefit of new technology.

3.2.3 Challenges of current solid waste management

During the last few years, the MSW quantity in Hoi An has been increasing significantly
due to tourism development activities. As a consequence, the capacity of collection,
transportation, treatment and disposal have become to be not sufficient. In addition, there is a
significant increase in the expense required for the collection, treatment disposal and treating
environmental pollution associated with MSW.

For instance, the investment for construction of transferring areas and the number of
transport vehicles, as well as storage equipment in public locations, has not met the increasing
demand for MSW. This leads to the less effectiveness of collection, separation at source as well
as accumulation of MSW in residential areas causing the odour problem. Moreover, collection
facilities and equipment has not met the technical standard and has not ensured the requirement
for environmental sanitation. Although the expenses used for collection, transportation has been
increasing but still not meet the requirement of the city development.

The recycle material activities have been contributing about 5% of MSW recycled (Chu,
2014), is carried out by informal sectors such as scavengers, recyclers and craft villages. Due to
lack of recycling technology, incentive policies and legal enforcements, the efficiency of
recycling activities is low, and it consists risks of pollution emission, disease exposures to the
community.

Moreover, inappropriate management of MSW also caused the negative community health
impacts. People living near the open dumping sites and unsanitary landfills are being exposed to
the adverse impacts with the symptoms of the dermatological disease, respiratory diseases, and
diarrhoea. The most vulnerable groups are scavengers on the site in which most of them are
women (D. L. Nguyen et al., 2013).

The open-dumped landfill has been a serious problem in the city not only in term of
environmental and community health impacts. Directly disposing of in an open site and poorly
63
managed manner have also caused social conflicts arisen between affected residence community,
waste management company and city authorities. Due to the failure of new treatment plants such
as composting and incinerators, residents have lost faith in waste managers and authorities. The
resident's belief in waste separation at source programme might be lost since they could not see
any improvement of the waste treatment practices and the environment. Lack of faith of citizens
could lead to the other failures in authorities efforts to improve waste management system in the
future.

Tourism has accounted for a significant amount of waste generation and has a strong effect
on MSW generation in Hoi An city (Hoang, Fujiwara, et al., 2016a; Hoang, Nguyen, et al., 2016).
Furthermore, environmental pollution associated with MSW management have resulted in
adverse effects on the tourism industry. Visibility of “mountain of waste” (open landfill), odour
pollution from waste collection and treatment site, waste left over on the tourist street due to the
weak of the collection system could leave the negative impression on tourists. The number of
tourists visiting the city could decline due to environmental problems, impact on economic
development.

The environment driver should be a major driver of waste management and disposal in Hoi
An city due to following reasons:

1) a clean environment can preserve human health, which is the first objective as well as the
indicator of the success of a waste management system;

2) environmental protection priority requires strong decisions from authorities, which might
motivate other drivers such as institutional, economic and human drivers.

For instance, decline the amount of waste landfilled by having an appropriate treatment
technology (such as incinerator) is essential to reduce contamination and environmental pollution.
That requires significant investment in terms technology and equipment (economic drivers). Also,
it encourages scientific research (institutional drivers) to develop local technology or tailor the
technology transferred from developed countries according to local conditions. Lack of
involvement of scientific research might be a reason for the failure of the current composting
plant and incinerator in Hoi An. Residents might have lost faith in authorities and the local
legislation due to the worsening of environmental conditions and vice versa (human drivers).
64
 Another driver should be considered as a significant driver for Hoi An is institutional drivers
including scientific research and legislation. Sustainable waste management is a long-term target
which can balance the three targets: environmental effectiveness, economic affordability and
social acceptance. Deciding the point of balance requires a comprehensive and precise research
based on natural, geographical and socio-economic conditions of target city.

In addition, local legislation can change the way waste materials are managed or disposed of
(P. Agamuthu et al., 2009). For example, in Hoi An, appropriate legislation to control waste from
tourism sectors (from hotels, restaurants, tourist places, etc.) could encourage MSW reduction
because of a significant contribution of waste from this area.

A decentralised treatment approach could benefit for degradable waste due to the failure of
the centralised treatment facility and the small amount of waste generation in further areas.
However, the approach needs to be studied carefully to have the applicable implications and
appropriate legislation to encourage decentralised waste treatment. Education and awareness
were implemented successfully in Hoi An and should be maintained and improved.

3.3 Research proposed for the case study

Persistent increase in waste generation coupled with highly variated waste composition due
to the impact of tourism has required a sustainable waste management in Hoi An city. It is not
only to ensure environmental protection and public health risk prevention but also the incentive
of tourism promotion and economic development. Several essential drivers being responsible for
a sustainable waste management in developing countries were discussed in previous studies (P
Agamuthu et al., 2007; P. Agamuthu et al., 2009; Wilson, 2007). However, the groups of drivers
work in different degrees in different times to affect the efficiency of a waste management, and
sustainable waste management is a long-term objective. Therefore, choosing balance point of
three factors: economic, environment and society in the first period of establishing a sustainable
waste management system is increasingly important.

Therefore, in this dissertation, a systematic approach which includes critical stages of waste
management planning is proposed to develop a sustainable waste management system.

The main objective of this study is to provide detailed information related to waste flow
management and appropriate treatment technologies options based on optimization analysis to
65
support decision makers in the decision-making process to establish a sustainable waste
management system for the city. To achieve this important purpose, the author also conducts
various analysis which is crucial for waste management modelling and planning such as
sampling and surveying, waste characterization analysis, waste generation modelling,
optimization modelling, creating a framework for discussion of decision makers.

The reliable data of MSW generation and composition is necessary for developing waste
management model as well as developing a 3R approach. Thus, the author will present the
procedure of waste data collection and analysis thoroughly to obtain reliable information about
waste generation rate and detailed composition which will be used as the input parameter of the
waste management model. Figure 3.6 presents the objectives and content of waste data sampling
and survey proposed

The author will evaluate the correlation between socioeconomic and demographic factors
with waste generation to provide supporting information for improvement of waste collection
and management system. The study will also develop a mathematical model for estimating
household waste generation. Also, various theoretical tests, as well as validation technique, will
be carried out for evaluating the performance of the model and choosing the “best” model for
predicting waste generation.

In addition, the aim of this study will describe and assess the current status of waste
management system, and identify the problems as well as discussing the solution. Thus, a single
objective model of waste management system will be developed to minimizing cost and
evaluating the initiatives of sustainable waste management. A comparative analysis of 12
scenarios with different waste to landfill target and waste separation efficiency initiatives will be
conducted to assess the waste stream management strategies and treatment alternatives based on
various aspects, such as GHG emission, energy generation/consumption, economic cost/benefit,
and land use burden.

66
 Figure 3.6. Targets and contents of waste data collection
It a heavy tasks to identify the balance point of three aspects of sustainable waste
management for decision makers. A multi-objectives optimization model which considers to
minimizing pollutants emission, minimizing cost and waste-to-landfill will be developed for
seeking feasibly optimal solution of waste stream management and treatment alternatives of
sustainable waste management. The mode proposed will overwhelm the limitation of previous
models by directly calculate various pollutants emission and considering all technologies’
emission such as composting, AD, landfill, etc.

Lastly, the social factors of sustainable management in the waste model will be improved in
this study by the involvement of waste generators and stakeholders in the decision-making
process. The reference point method is applied for solving the optimization model to get the
optimal solution for sustainable waste management. A framework of procedures for different
DMs work and discuss to get an agreement on a compromised solution for waste management is
introduced.

3.4 Reference
Agamuthu, P., Fauziah, S., Khidzir, K., & Aiza, A. N. (2007). Sustainable waste management-
Asian perspectives. Paper presented at the Proceedings of the international conference on
sustainable solid waste management.
Agamuthu, P., Khidzir, K. M., & Hamid, F. S. (2009). Drivers of sustainable waste
management in Asia. Waste Management & Research, 27(7), 625-633.
doi:10.1177/0734242x09103191

67
 ASTM-E897-88. (2004). Standard Test Method for Volatile Matter in the Analysis Sample of
Refuse-Derived Fuel (Withdrawn 2004). West Conshohocken, PA, 1988: ASTM International.
ASTM-E949. (2004). Standard Test Method for Total Moisture in a Refuse-Derived Fuel
Laboratory Sample (Withdrawn 2004). West Conshohocken, PA, 1988 ASTM International,.
ASTM-E1109. (2009). Standard Test Method for Determining the Bulk Density of Solid Waste
Fractions. West Conshohocken, PA, 1987: ASTM International.
Chang, N.-B., & Pires, A. (2015). Systems Engineering Tools and Methods for Solid Waste
Management Sustainable Solid Waste Management (pp. 235-299): John Wiley & Sons, Inc.
Chu, M. T. (2014). Báo cáo tổng kết đề tài: Xây dựng mô hình đồng quản lý rác thải tại hai xã,
phường Cẩm Hà và Cẩm Phô, thành phố Hội An (Project report: building co-operative waste
management model in two wards, Cam Ha and Cam Pho, Hoi An city). Retrieved from Hoi An,
Viet Nam:
Denafas, G., Ruzgas, T., Martuzevičius, D., Shmarin, S., Hoffmann, M., Mykhaylenko, V., . . .
Ludwig, C. (2014). Seasonal variation of municipal solid waste generation and composition in
four East European cities. Resources, Conservation and Recycling, 89, 22-30.
doi:http://dx.doi.org/10.1016/j.resconrec.2014.06.001
E711-87, A. (2004). Standard Test Method for Gross Calorific Value of Refuse-Derived Fuel
by the Bomb Calorimeter (Withdrawn 2004). West Conshohocken, PA, 1987: ASTM
International.
Espinosa Lloréns, M. d. C., Torres, M. L., Álvarez, H., Arrechea, A. P., García, J. A., Aguirre,
S. D., & Fernández, A. (2008). Characterization of municipal solid waste from the main landfills
of Havana city. Waste Management, 28(10), 2013-2021.
doi:http://dx.doi.org/10.1016/j.wasman.2007.07.004
Faraway, J. J. (2005). Linear models with R. Boca Raton, FL. [etc.]: Chapman & Hall/CRC.
HAPWS, J. (2014). Báo cáo công tác quản lý chất thải rắn tại thành phố Hội An (Report on
solid waste management of Hoi An city). Retrieved from Hoi An, Vietnam:
HAPWS, J. (2015). [Báo cáo công tác quản lý chất thải rắn tại thành phố Hội An (Report on
solid waste management of Hoi An city)].
HASD. (2013). Statistical Yearbook - Hoi An city. Hoi An Statistical Department, Viet Nam.
Hoang, M. G., Fujiwara, T., & Pham Phu, S. T. (2016a). An investigation of household waste
generation and composition in Hoi An, Viet Nam. Paper presented at the The 3rd International
scientific conference on Material Cycles and Waste Management (3RINCs) Hanoi, Viet Nam.
Hoang, M. G., Fujiwara, T., & Pham Phu, S. T. (2016b). Municipal waste generation and
composition in a tourist city - Hoi An, Vietnam. [Manuscript has been accepted for a journal.].
Hoang, M. G., Nguyen, T. V. P., & Fujiwara, T. (2016). Feasibility Study on Effectiveness of
Solid Waste Incinerator in Hoi An, Vietnam. Paper presented at the Paper presented to the 9th
International Conference on: Combustion, Incineration/Pyrolysis, Emission and Climate change
(the 9th i-CIPEC) Kyoto, Japan. http://epsehost.env.kyoto-u.ac.jp/i-cipec2016/

68
 Longnecker, M., & Ott, R. (2001). An introduction to statistical methods and data analysis.
ISBN-13, 854576151.
Madigan, D., & Raftery, A. E. (1994). Model selection and accounting for model uncertainty in
graphical models using Occam's window. Journal of the American statistical Association,
89(428), 1535-1546.
Nguyen, D. L., Hoang, M. G., & Bui, X. T. (2013). Challenges for municipal solid waste
management practices in Vietnam. Waste Technology, 1(1), 17-21.
Nguyen, T. K. T. (2014). Municipal solid waste management in Vietnam challenges and
solutions Municipal Solid Waste Management in Asia and the Pacific Islands (pp. 355-377):
Springer.
Nordtest. (1995). Solid waste, municipal: Sampling and characterisation (NT ENVIR 001).
Nordtest method (Publication no.
http://www.nordtest.info/index.php/methods/environment/item/solid-waste-municipal-sampling-
and-characterisation-nt-envir-001.html?category_id=24). http://www.nordtest.info
Perny, P., & Weng, P. (2010). On Finding Compromise Solutions in Multiobjective Markov
Decision Processes. Paper presented at the ECAI.
Scheaffer, R., Mendenhall III, W., Ott, R., & Gerow, K. (2011). Elementary survey sampling:
Cengage Learning.
Sfeir, H., Reinhart, D. R., & McCauley-Bell, P. R. (1999). An evaluation of municipal solid
waste composition bias sources. Journal of the Air & Waste Management Association, 49(9),
1096-1102.
Shamshiry, E., Nadi, B., Bin Mokhtar, M., Komoo, I., Saadiah Hashim, H., & Yahaya, N.
(2011). Integrated models for solid waste management in tourism Regions: Langkawi Island,
Malaysia. J Environ Public Health, 2011.
Teh, L., & Cabanban, A. S. (2007). Planning for sustainable tourism in southern Pulau Banggi:
an assessment of biophysical conditions and their implications for future tourism development.
Journal of Environmental Management, 85(4), 999-1008.
Wierzbicki, A., Makowski, M., & Wessels, J. (2000). Model-based decision support
methodology with environmental applications: Kluwer Academic Dordrecht, The Netherlands.
Wierzbicki, A. P. (1982). A mathematical basis for satisficing decision making. Mathematical
Modelling, 3(5), 391-405. doi:http://dx.doi.org/10.1016/0270-0255(82)90038-0
Wilson, D. C. (2007). Development drivers for waste management. Waste Management &
Research, 25(3), 198-207.

69
4. MUNICIPAL SOLID WASTE CHARACTERIZATION

4.1 Introduction

Rapid urbanisation and tourism development have led to significant increase in municipal
solid waste (MSW) generation in Hoi An City (HAC), a famous tourism city the centre of
Vietnam. Currently, improper disposal (an open-dumped landfill) and not effective treatment
method (a composting plant) in the city have caused adverse impacts on all components of the
environment and human health. Therefore, MSW management is one of the most significant
environmental problems in Hoi An and developing a solid waste management system with
material and energy recovery capabilities has become a necessity in the City.

It is clear that understanding the features of the waste stream is crucial for planning, design,
and operation of a solid waste management (SWM) system (Abu Qdais et al., 1997; Chang et al.,
2015). To develop a waste management system and choose appropriate treatment technologies,
accurate and reliable data on the waste composition and generation are needed (Edjabou et al.,
2015). Also, understanding physical properties and chemical composition of MSW components
is crucial in evaluating treatment options. However, physical, geographical, sociocultural,
economic, and political factors may have influences on the composition and generation of
municipal solid waste (MSW) (Gallardo et al., 2014; Gidarakos et al., 2006). Especially, the
tourism industry may lead to significant influences to the environment with respect to resource
consumption as well as pollution generation. One of the major impacts of tourism is the change
in generation rate and composition of MSW as well as household waste. Some previous studies
have reported the increase in MSW in the tourist areas resulting from the high number of tourists
during the tourism season (Denafas et al., 2014; Espinosa Lloréns et al., 2008; Shamshiry et al.,
2011; Teh et al., 2007). Thus, waste composition and waste generation rates can be different in
places influenced by, and in those not influenced by tourism. Study on variations of waste
characterization among various regions in a tourist city may be beneficial in planning a
sustainable system.

Study on waste has still lagged behind where it concerns waste arising from tourism
operations due to the near absence of reliable data on MSW management related to this sector

70
(Ezeah et al., 2015). In HAC, there are no studies on waste characterization in the city; a waste
composition monitoring program has been carried out recently, but there has been no information
on waste generation and waste composition from different sectors. The tourism industry in HAC
now is considered one of the most dynamic industries and the greatest contributor to the
development of the city; hence, a study on waste characterization of household and tourism
sectors from various geographical areas would have a great contribution to a sustainable MSW
management system planning.

The aim of this study was to provide detail waste generation and composition to evaluate
differences in waste characterization among different types of urban areas as well as between
household and tourism, the two biggest sources of municipal solid waste in the city. Results of
the study would play a major role in developing a sustainable system such as choosing
appropriate treatment options, designing collection system, or operating solid waste management
system in the city. We also provided an advantageous framework including sampling procedures
and appropriate statistical analysis, as a reference that is readily applicable to small-scale cities in
Vietnam.

4.2 Methodology

4.2.1 Research target areas and number of samples

HAC comprises 13 administrative wards including an island ward that has its old waste
management system. Thus, in this study, we focus on 12 neighbourhoods in the mainland. The
City is a unique urban area in Vietnam, which consists of many types of areas including urban,
rural places, and a famous tourist quarter (Hoi An ancient town). Therefore, in consideration of
tourist factors affecting household waste generation, we divided the city into three strata (ST).
ST1 is the Hoi An ancient town area including Minh An ward and part of Cam Pho Ward; ST2
comprises eight urban and suburban wards, and ST3 is a group of the other countryside wards
(Tan Hiep island is excluded in this study). Figure 4.1 describes the three strata in this study; the
ST1, ST2, and ST3 areas are represented by red, green, and blue, respectively.

The sampling strategy adopted in this study is stratified random sampling to reduce
variations in household waste. We can apply the graph of Van der Broek and Kirov (Van der
Broek et al., 1971) (Figure 4.2), which showed that the sampling ratio should be about 13
71
households per 1000 households to obtain a standard of error of 5% (Nordtest, 1995). The
number of statistical samples was estimated based on the data of households number in each

Figure 4.1. Target research areas.

ward from the Hoi An statistical yearbook 2013 (HASD, 2013). Households participating in the
survey program were randomly chosen from each stratum by R software from the lists of
households in three strata, gathered from the Hoi An Center for Population and Family Planning.

Source: (Nordtest, 1995)

Figure 4.2. Sampling ratio

72
 To determine the waste characterization from various sources of the city, not only waste
from households but also waste from tourist sources including hotels (HT), restaurants (RBC),
and streets (STR) of tourist corners was chosen as targets of the study. The Hoi An Statistical
Yearbook (2013) was used to identify and shortlist hotels and restaurants in the city, which
would be chosen randomly as samples for waste quantity sampling. In Hoi An, there are several
kinds of accommodations such as hotels, homestays, and guesthouses. Totally, nine types of
accommodations for tourists including five hotels, three homestays and one guesthouse were
selected randomly.

Table 4.1. Numbers of statistical samples adopted in the survey program.

Code Type of Sources No. of housholds Number of samples (n)

(HASD, 2013) Statistic required Number of samples
ST1 Stratum 1 2.752 36 50
ST2 Stratum 2 14.394 185 187
ST3 Stratum 3 4.904 64 84
ST Total households 22.530 285 321
HT Hotels - 9
RBC Restaurants - 6
STR Streets - 3
For selecting restaurants, as suggested by independent local experts and authorities,
restaurants in Hoi An were sorted into two types based on their locations: in the city center and
along the beach. Thus, three of each type were randomly chosen for the survey. Likewise, three
walking streets in the Hoi An ancient quarter were selected to identify the amount of street waste
generation and composition. In this study, we randomly chose nine hotels, six restaurants, and
three walking streets in tourist corners for samples collection. The number of households, hotels,
restaurants, and streets in the survey is described in Table 4.1.

4.2.2 Sampling procedures

The period of sampling was three weeks from August 2 to August 20, 2015. On the first
week, we focused on training both students engaging in the sampling program and families
selected as donors of household waste. The actual sampling time was 14 days from August 7 to
August 20. Every day, we collected wastes in the morning from households, hotels, restaurants

73
and streets, then transported them to the waste treatment plant and separated them for
composition analysis.

There were 25 students recruited from Danang College of Technology (DCT) in Vietnam.
These students were trained and provided technical, safety guidelines, as well as personal
protecting equipment such as rubber gloves, safety glasses, masks, and other equipment
including plastic buckets, handkerchiefs, and record paper sheets. In the first stage, all groups
went to visit every single household invited to participate in the program and collected basic
information (e.g., number of household members). As a result, a total of 321 households agreed
to participate in the survey of waste sampling collection. Total households for each stratum were
50 for Stratum 1, 187 for Stratum 2, and 84 for Stratum 3.

Each family had a single identified code including two parts: the first part consisted of three
digits presenting the name of the stratum (ST1, ST2, ST3) and the second part was the number of
household by numerical order starting from 1 in each stratum. Every single household
participating in the program was assigned a code marker in front of the door to avoid collection
mistakes. The procedures of collecting, marking and analysis waste samples is presented in the
figure 4.3.

Figure 4.3. Waste sampling procedure

Waste sample collection was carried out from each selected household daily for two weeks
from August 7th to August 20th. On the first day, not all waste samples were collected, while the
second-day waste was collected completely but was later discarded to ensure that the waste to be
analyzed had been generated in the last 24 hours. Every morning, collecting team members
brought black plastic bags with the sample identification codes and transported the waste bags
with household waste to the treatment plant to measure and record the weight.
74
 Samples were separated few hours after being collected to minimize variations due to
biological and chemical processes. On the first day of waste composition analysis, bags of
household waste were randomly selected from those collected so that the total weight of the bags
may be around 120 kg. Thus, 70 waste bags from 70 households were selected randomly as
samples for composition analysis including 13 from Stratum 1, 36 from Stratum 2, and 21 of
Stratum 3. Samples of household waste that had been chosen on the first day would be kept as
samples for the whole survey period.

On the other hand, bags of hotels and restaurants waste were selected randomly for
composition analysis every day due to the huge amount of waste collected daily from these two
sources. There was an average of around 105 kg of household waste and 145 kg of waste from
restaurants, hotels, and streets separated daily. Waste samples were separated into ten primary
components presented in Table 4.2. In this study, the effects of seasonal variations in waste
characteristics were not investigated.

Table 4.2 Categories of waste separation in samples

Primary Material
components
Food waste Degradable food remnants
Yard waste Leaves, residual from gardens, etc.
Plastic Plastic bags, plastic bottles, PET bottles, or plastic packages, etc.
Cardboard Cardboard, packing cardboard, cartons, milk and juice package
Paper Newsprint, magazines, office paper, mixed paper, tickets
Metal Metal packaging, non-packaging metal, aluminum, iron, ferrous metals,
non-ferrous metals, etc.
Glass Glass bottles, all kinds of glasses, etc.
Combustible Textiles and leather, healthcare textiles, carpeting and mats, wood,
bamboo, rubber, napkins, sanitary napkins, etc.
Incombustible Inert materials and non-combustible materials, etc.
Hazardous waste Battery, medical waste, electrical devices, etc.

75
4.2.3 Physical and chemical analysis

In this research, we analysed the waste characteristics including specific weight (ASTM-
E1109, 2009), moisture content (ASTM-E949, 2004), volatile matter and ash content (ASTM-
E897-88, 2004), energy content (E711-87, 2004) and chemical elements (C, H, N) (using CHN
analyser). We collected 37 samples after composition analysis for laboratory analysis; the
samples selected did not include non-combustible materials. For each category such as Food,
Garden, Plastic, Cardboard, Paper, Combustible and Hazardous waste several samples were
collected randomly for analysis.

Waste density was analysed right after waste composition analysis; waste samples were
dried in the dryer for 48 hours (for moisture content analysis) then were taken to the laboratory
for further analysis.

The energy content of municipal solid waste and C:N ratio of Food waste is measured, we
assumed inorganic carbon of Food waste equals zero. The statistic calculation for the mean of the
component and analysed values is calculated as (ASTM-D5231-92, 2016).

Figure 4.4 Physical and Chemical analysis

4.2.4 Statistical analysis

Daily waste generation of household j in strata i is expressed as wi,j. Waste generation rate
and its average in strata i are expressed in Eq (4.1) and Eq (4.2), respectively.

wi , j
yi , j  (4.1)
hi , j

76
 ni
1
yi 
ni
y
j 1
i, j
(4.2)

Where hi,j is the number of people of the household j in stratum i; yi,j is the daily waste
generation per capita of household j in stratum i; and ni is the number of sample households in
stratum i.

In this study, we used geometric and standard deviation (Tchobanoglous et al., 1993) for
statistical measures to describe the mean of waste generation and percentages of waste
components. For stratified random sampling, population mean of waste generation and
composition was unbiasedly estimated by a weighted average of means of strata (Bolaane et al.,
2004; Scheaffer et al., 2011), described by the following equation:

1 L
y st 
N
n y
i 1
i i (4.3)

N, L is the total number of households being surveyed and the number of strata, respectively.

Distribution of waste generation rate for each stratum is examined by using skewness and
kurtosis. Skewness, expressed by Eq (4.4), is a measure of the asymmetry of the probability
distribution of a real-valued random variable about its mean. The skewness value can be positive
or negative, or even undefined. Kurtosis, expressed by Eq (4.5), is a measure of whether the data
are heavy-tailed or light-tailed about to a normal distribution.

1 Ni
y  yi 
3


i, j
Si  3
(4.4)
Ni j 1 si

1 Ni
y  yi 
4

Ki 
Ni
j 1
i, j

si
4
(4.5)

Where si is a standard deviation of yi,j.

An analysis of variance (ANOVA) was carried out to analyze the significant differences in
mean waste generation rate and waste compositions among sources and geographical areas. The
total waste generation data and quantity of waste types were entered into Microsoft Excel 2013
to estimate means of waste generation per capita (for households), per room (for hotels), and 100

77
meters (for street) and percentage of waste components. Results were transformed by taking the
logarithm of their original values (Bernache-Pérez et al., 2001; Brown et al., 2002) to
approximately close to a Normal distribution. Then, they were used as input data for R software
(ver 3.2.0) where ANOVA was tested. To compare between strata, a post-hoc analysis was
applied with Turkey’s Honest Significance Differencemethod (Longnecker et al., 2001; Tuấn,
2014) for pair comparisons.

During the time of survey program, waste collection from 30 households was not consistent.
Therefore, we disregarded the ddata gathered from those families in the statistical analysis. As a
consequence, data of 291 statistical samples for households were selected for analysis.

4.3 Result and discussion

4.3.1 Municipal waste generation

4.3.1.1 Household waste generation

The waste generation rate of each household was calculated by dividing the amount of
waste produced in a day with the number of family members (Dangi et al., 2011). Figure 4.5
shows the results of statistical estimates and the boxplot of waste generation rate; red dots
present the mean of waste generation rate of each stratum, and small black dots are outlier
observations.

Waste generation of ST1 and ST2 ranged from 0.00 to 0.70 kg capita-1 day-1, but ST3 took a
lower range of 0.00 - 0.30 kg capita-1 day-1. These results showed that waste generation of
households from rural areas (ST3) varied less than from urban areas (ST1 and ST2).

Table 4.3 Means of waste generation and 95% confident interval.

Stratum Waste generation rate (kg capita-1 day-1)

Mean Upper limit Lower limit
ST1 0.203 0.046 0.900
ST2 0.264 0.085 0.823
ST3 0.120 0.024 0.604

78
Figure 4.5 Boxplot of waste generation rate Figure 4.6 Density of waste generation rate

Figure 4.6 shows the density of waste generation rate of each stratum. It was found that each
waste generation distribution of the three strata positively skewed like a normal bell-shaped
curve distorted to the left. According to Tchobanoglous et al. (1993), some degree of positive
skewness is common in solid waste generation data. The ST3’s distribution peaked because the
coefficient of kurtosis was higher than the coefficient of kurtosis of the normal distribution,
which was equal to 3 (Bolaane et al., 2004; Dangi et al., 2008; Tchobanoglous et al., 1993).
Previous studies used mean estimates and standard deviation (SD) (Bolaane et al., 2004; Gomez
et al., 2008; Philippe et al., 2009; Qu et al., 2009; Thanh et al., 2010) to describe the confidence
interval of waste generation rate by tolerating the small amount of skewness to have Normal
distribution (Dangi et al., 2008). However, in this study we used geometric mean and standard
deviation for statistical estimates of waste generation rate (Tchobanoglous et al., 1993). The
geometric mean of waste generation rate and 95% confidence interval are shown in Table 4.3. A
person living in the countryside (ST3) generated half the amount of waste produced by a person
living in the urban area. Among different waste generation studies at household level in Vietnam,
the rate was estimated at 0.285 kg capita-1 day-1 in Can Tho City (Thanh et al., 2010) and 0.6 in
Ho Chi My City, as reported by Tran et al. (Tran et al., 2014). Both cities mentioned above are
bigger and more developed than Hoi An. Thanh et al. (Thanh et al., 2010) sampled from 100
households for both rainy and dry seasons then estimated the mean of two results. Tran et al.
(Tran et al., 2014) meanwhile, combined the amount waste after household separation to estimate
79
the daily waste generation per household. This study assumed the average number of a family
member was four to calculate waste generation per capita. Results of this study showed that the
household waste generation rate in Hoi An was slightly lower compared to those in Can Tho city
and Ho Chi Minh City. The overall mean of household waste generation rate was0.223 kg capita -
1
day-1 with average varying between 0.207 and 0.240 kg capita -1 day-1. Thus, total household
waste generation of the city was estimated at 20.75 tonnes per day.

Table 4.4 Results of variance analysis

Source DOF1 SOS2 MS3 F4 P5

Between Strata 2 34.7 17.35 37.5 3.4.10-15
Within Strata 288 133.3 0.46
1
Degree of freedom; 2 Sum of square; 3 Mean square; 4 F-value; 5 P-value

Figure 4.7 presents the post-hoc analysis results. Horizontal lines indicate the 95%
confidence interval of differences of average waste generation between strata. If the 95%
confidence interval line is not crossing the zero line, it meant waste generation rate was
significantly different between two strata. The result showed that waste generation rate in ST3
was significantly different from those in ST1 and ST2. Thus, the waste generation rate from
urban areas was double that from rural places, and the difference was statistically significant.
The result of the study would be useful in the prediction of waste generation for a waste
management planning program.

Figure 4.7 95% family-wise confidence level

80
 Tourist area (ST1) produced less waste per capita than did the urban and suburban areas
(ST2) as a result of this survey. Family business and involvement in tourism services of people
living in the center (ST1) had significant impact on family lifestyle, such as reduced time at
home, eating outside, or less time spent together with family members could be a reason for less
waste produced. However, this difference was not statistically significant as shown in Figure 4.7.

According to the data of Hoi An Public Works Service Co. Ltd. (HAPWC) report regarding
the daily waste amount, we estimated the average amount of city municipal waste transported to
a treatment facility, which was about 70.13 tonnes day-1. Therefore, daily household waste
generation accounted for approximately 30% of total municipal waste in Hoi An City. The
results of ANOVA (Table 4) indicated that there were differences between waste generation
rates among the three Strata (signals were much higher than noises), and the differences were
statistically significant (p-value<0.05).

4.3.1.2 Waste generation from tourism sectors

The estimated average waste generation from hotels was 1.22 kg room -1 day-1. However,
there was a significant difference between hotels with large capacity (four stars hotels in this
study) and hotels with small capacity (three stars and two stars hotels and homestays.

As can be seen from Figure 5, high standards hotels generated about 4 kg room -1 day-1;
meanwhile, three stars and lower standard hotels produced about 0.35 kg room -1 day-1. Chan and
Lam(Chan et al., 2001) reported that mean of waste generation from a room in midscale hotels
(which included three and four stars hotels) in Hong Kong was 2.76 kg room -1 day-1 and from
hotel restaurants were 0.75 kg per meal cover in 1996. The hotel waste generation rate tended to
decrease as a result of Chan and Lam(Chan et al., 2001) estimated model. However, waste from
guest rooms and restaurants of hotels were studied separately in the research mentioned above.

The results of this study were significantly lower than those of Trung et al. (2005), who
estimated waste generation rate by a questionnaire survey of 50 hotels in Vietnam. Trung et al.
(2005) reported that two and three stars hotels generated 0.7 – 5.6 and 8.2 – 17.9 kg guest-1 day-1,
respectively while four stars hotels produced 13.5 – 32.3 kg guest-1 day-1. Byer et al. (2006)
meanwhile, indicated that the waste generation rate of two, three, and four stars hotels appeared
identical from 0.4-0.5 kg guest-1 day-1.
81
 (a) (b)

Figure 4.8 Boxplot of waste generation from tourism sector

Figure 4.9 describes the correlation between the number of hotel rooms and waste
generation rate. The results of correlation analysis showed that the coefficient of correlation (r)
was 0.95, and it was statistically significant (p<0.05). The coefficient of determination was 0.91,
which meant that 91% variations in waste generation rate could be explained by the number of
hotel rooms.

r = 0.95

R2
= 0.91

Figure 4.9 Correlation of mean of waste generation and number of hotel room

82
 The average waste generation measured was 28.85 kg day-1 from restaurants and 7.35 kg
100m-1 day-1 from street litter. On 95% Confidence interval, the means of waste generation rate
from tourism sources were estimated as follows:

Hotels: 0.6 kg room-1 day-1, range from 0.05 to 6.07.

Restaurants: 26.18 kg day-1, range from 7.68 to 89.30.

Streets: 6.99 kg 100m-1 day-1, range from 3.56 to 13.67.

4.3.2 Municipal waste composition

4.3.2.1 Household waste composition

The composition of household waste was made up of 38% food waste, 19% yard waste,
14% plastic, 15% combustible waste, and others including cardboards, paper, metals, glass,
incombustible waste, and hazardous waste constituted less than 5% (Figure 4.10). The major
components of household waste in Hoi An were food waste and yard trimmings (19%), which
together constituted 57% of the total household waste. Metal and hazardous material were the
smallest contributors to household waste stream.

Figure 4.11 provides the waste compositions of households from the individual stratum. In
comparison between strata, the portion of food waste in ST1 was higher (at about 43.3% on
average) but less garden waste (7.54%) than the other two strata. ST2 and ST3 had nearly similar
percentages of food and garden waste of around 35% and 19%, respectively. The percentages for
most of the categories of ST2, ST3 (Fig.8), and total household (Fig.7) were apparently identical.
Meanwhile, ST1 (Fig.8) and Tourist (Fig.7) waste were similar in shape to a higher percentage of
food and glass waste, though a lower proportion of garden waste. Thus, the household waste
characterization was different between an area affected by tourism (ST1), and two areas less
affected by tourism activities (ST2 and ST3). However, the results from ANOVA indicated that
differences in waste composition were not statistically significant (p<0.05).

83
Figure 4.10 Waste composition of household Figure 4.11 Household waste compositions of
and tourism sector. thee strata.

4.3.2.2 Tourism waste composition

The waste stream of tourism sources (including hotels, restaurants, and streets) was different
compared to that of the household waste stream (Figure 4.10 and Table 4.5). The portion of food
waste of tourism was slightly higher (42%) but the garden waste was significantly lower (4.5%)
than that of household. Glass materials released from the tourism industry were far greater than
those from households. It was understandable because tourism service activities produced huge
amounts of glass, especially beverage bottles. Both household and tourism wastes had high
portions of plastic and combustible materials (as explained in Table 4.2).

Table 4.5. shows the percentage of waste compositions from hotels, restaurants, and streets
separately. All three sources of waste had a high portion of food waste. Activities of restaurants
on streets and leftovers discharged by walking tourists could explain the high amount of food
waste in street litter bins. Naturally, we found higher amount of glass waste such as discarded
beer and wine bottles during the sampling period.

84
 Table 4.5 Waste composition of different sources regarding tourism waste

Waste Compositions (%) Hotel waste Restaurant waste Street waste

Food 49.0 39.5 39.4
Garden 7.2 6.4 1.4
Plastic 16.4 8.4 13.8
Cardboard 3.2 1.9 1.0
Paper 9.7 9.6 6.7
Metals 0.7 2.2 0.4
Glass 3.4 9.4 4.5
Combustible 8.6 15.3 31.8
Incombustible 0.7 7.0 0.8
Hazardous 1.0 0.3 0.2

Chan and Lam (Chan et al., 2001) declared that plastic toiletries constituted the highest
portion (above 50%) of waste from hotel guest rooms. Byer et al.,(Byer et al., 2006) indicated
that compostable material (including food waste and yard waste) formed the largest portion of
the hotel waste stream, accounting for approximately 60%-70%. It was slightly higher than the
compostable material portion in this study.

4.3.2.3 Municipal waste composition

Figure 4.12 Municipal waste compositions

85
 Figure 4.12 shows the municipal waste composition of the city, in which the largest
proportion was food wastes (42%) and hazardous waste was the least portion (less than 1%).
Garden waste, plastic, and combustible materials were significant contributors ranging from 12%
to 16%. Other components were less than 10% of total waste. Similar values of total degradable
waste were reported by Hoi An Public Works Service Co. Ltd. (HAPWC) in September 2015.
However, the report indicated that the percentage of yard waste was much higher than that of
food waste. Figure 4.12 presents the percentage of waste sub-categories

Studies for solid waste composition analysis implemented various approaches without
common international standards. Therefore, it was difficult to compare results of different
sampling and sorting methodologies (Dahlén et al., 2008; Edjabou et al., 2015). The portion of
biodegradable waste (food waste and garden waste) was slightly lower than those in developing
cities such as Beijing, China (69.3%) (Qu et al., 2009), Cape Haitian, Haiti (66.6%) (Philippe et
al., 2009), Kathmandu, Nepal (71%), Can Tho, Vietnam (80%)(Thanh et al., 2010) and Da Nang
City, Vietnam (68,5%)(MONRE, 2011). However, recyclable waste components such as plastics,
cardboards, papers, and metals were approximately alike compared to the above studies. The
difference could be caused by different lifestyles and dietary habits as well as differences in the
effects of tourism industry between Hoi An and other cities. Combustible waste, on the other
hand, was not investigated in the studies mentioned above, but it was one of the major
components in municipal waste in HAC.

Figure 4.13 Hoi An MSW composition 18 sub-categories

86
4.3.3 Waste characteristics

The results of MSW density analysis are shown in Table 3. The mean specific weight of Hoi
An MSW is 263 kg/m3. Household waste density is about 226 kg/m 3, lower than Hotel,
restaurant waste and street waste. Density of street waste is higher (317 kg/m 3) than waste from
other sources.

Table 4.6 Waste characterization in Hoi An city

Sources Waste density (kg/m3)

Foo Gar Pla Plb Car Pap Met Gla Com Int Haz Total
Household 343 118 29 55 712 88 231 208 154 527 176 226
Hotel, Restaurant 388 116 34 49 672 114 250 363 245 325 327 281
Street 377 91 31 57 613 99 600 811 315 921 1,050 317
MSW 368 113 32 53 680 101 294 366 218 495 372 263
C:N ratio 21.5 - - - - - - - - - - -
Foo: Food; Gar: Garden; Pla: Plastic; Plb: Plastic bag; Car: Cardboard; Pap: Mixed paper; Gla: Glas and Ceramic; Int: Inert material; Haz:
Hazardous material

Figure 4.14 and Figure 4.15 present the results of MSW physical, chemical characteristics
analysis and LHV measurement. The moisture of MSW is about 50%, Ash and Carbon content is
approximately 10% and 40% approximately.

Figure 4.14 Physical and chemical characteristics of MSW in Hoi An city

87
 The moisture content of MSW in Hoi An is about 53%. The high moisture content is caused
by Food and Garden waste, which accounted for 40% and 10% in composition and the moisture
content is 75% and 57% respectively. The high moisture will affect the thermal treatment of
MSW but is essential for biological treatment technology. Also, the C:N ratio of biodegradable
waste (Food and Garden waste) is 22.1, which is a good ratio for composting. However, since the
composting planting in Hoi An has been operating ineffectively, a deeper study in socio-
economical and technical efficient assessment might be needed to improve the treatment
efficiency.

Figure 4.15 Calorific value of MSW in Hoi An city

The heating value of MSW is about 9.4 MJ/kg waste. A In the waste management hierarchy,
waste incineration has been considered as a mode for energy recovery. As indicated above, the
average calorific value of Hoi An MSW is about 9.4 MJ/kg (ranged from 2 – 30 MJ/kg). s a
result of comparing the moisture, ash and carbon content to the Tanner triangle (WB, 1999),
incineration technology is a feasible option in term of waste combustibility. An incineration
could be a potential technology to reduce the waste amount quickly in the city. Especially, the
landfill of the city have been operating improperly and have been out of capacity.

88
 WTE have been one of the best technology in term of energy recovery in MSW management
(Kathirvale et al., 2004). The result from energy content of MSW in Hoi An indicated that a
WTE system might be a potential option for the city waste treatment system. However, the total
amount of waste produced is about 70 tonnes per day, the capacity of waste incineration will be
not high. Thus, cost-benefit study for incineration without energy recovery facilities should be
carried out.

Significant quantities of food waste, garden waste, and combustible waste proved that we
might consider introducing biological and thermal treatment technologies as potential treatment
options for municipal solid waste. However, the composting plant in Hoi An currently has been
operating unsuccessfully like most of the other composting plants in Vietnam. Failure in
composting technologies are caused by various factors such as inappropriate pre-treatment
techniques, not enough attention to the biological process requirements; poor quality of the
fertilizers, and poor marketing experiences (Nguyen, 2014). Centralized composting plant with
high capacity might be improper for the Hoi An condition due to lack of management skills,
operating funds, and low technology application. Thus, decentralized biological treatment
(small-scale plants, home composting) seem advisable to apply in these conditions, especially
with households in rural and suburban areas that have gardens and engaged in agriculture.
However, socio and economic incentives should be studied to encourage decentralized organic
waste treatment.

Thermal treatment, on the other hand, has been considered one of the best technologies in
term of energy recovery and reducing waste to landfill in MSW management (Kathirvale et al.,
2004). Waste incineration has been getting more attraction from the city authorities. The high
portion of combustible waste in Hoi An indicated that incineration technology might be a
potential option for the city waste treatment system. Nevertheless, the total amount of waste
produced is about 70 tonnes per day, and the capacity of waste incineration will be not high.
Thus, a study on the characteristics of waste such as heating value, moisture content, and
chemical elements (C, N, H) is required to determine the potential of incineration. A cost-benefit
study for incineration without energy recovery facilities also should be carried out.

89
 Recyclable waste (plastic, cardboard, paper, and metal) accounted for a large percentage of
around 20% giving a good chance for material recovery. However, informal recycling activities
for plastics, cardboard, and some other materials by organized waste pickers at treatment plants
are currently adequate. Also, current recycling activities employ residents living nearby Hoi An
waste treatment plant and are supported by local people. A material recovery program, therefore,
may not be warranted for now because of social and technological constraints, but should be
taken into account in the future.

4.4 Conclusion

Detailed survey and analysis of waste generation and composition in a tourism city, Hoi An,
Vietnam, were conducted to evaluate differences among types of urban as well as between
household waste and tourism waste. The findings are as follows:

Daily per capita household waste generation was 0.223 kg capita -1 day-1. Total household
waste generation rate in Hoi An City was 20.75 tonnes per day, which accounted for about 30%
of total municipal waste generation.

People living in rural areas generated about half of the amount of daily waste compared to
people living in urban areas. On the other hand, residents in the urban areas highly affected by
tourism activities produced less waste than in areas not affected by such activities.

Hotels generated about 0.6 kg room-1 day-1 and one restaurant in HAC produced an average
of 26.18 kg day-1. The mean of waste generation from tourist streets was 6.99 kg 100m -1 day-1.
The study found that hotels with garden and higher standard (four stars) created more waste than
lower-standard hotels. Regarding the correlations between the waste generation rate and the
number of hotel rooms, the result indicated a statistically significant correlation between the
number of rooms and total waste generation of hotels. The coefficient of correlation (r) was 0.95,
and the coefficient of determination was 0.91.

The municipal waste composition of the city had food waste as the largest proportion (42%)
and hazardous waste as the smallest contributor (less than 1%). Total biodegradable waste (food
and yard trimmings) was approximately 53%. Combustible waste was the second significant

90
component of about 16% while other recyclable contributed about 20% of the municipal waste
composition.

Waste composition in the city center was affected by tourism services because it was
different from other areas and likely alike with the waste composition of the tourism sector in
food, garden, and some other components.

Results from this study suggested a methodology of waste characterization survey based on
characteristics of urban areas. The results and methodology are accepted to be informative for
authorities, decision makers, stakeholders, and planners in developing a waste management plan.

Regarding the potential treatment technologies, we suggested that biological and thermal
treatment technology should be considered as highly potential options for municipal solid waste
to reduce waste to landfill. A recyclable material program should be developed in the future
toward sustainable waste management.

4.5 Reference

Abu Qdais, H. A., Hamoda, M. F., & Newham, J. (1997). Analysis of Residential Solid Waste
At Generation Sites. Waste Management & Research, 15(4), 395-405.
doi:10.1177/0734242x9701500407
ASTM-D5231-92. (2016). Standard Test Method for Determination of the Composition of
Unprocessed Municipal Solid Waste. West Conshohocken, PA, 2016: ASTM International.
ASTM-E897-88. (2004). Standard Test Method for Volatile Matter in the Analysis Sample of
Refuse-Derived Fuel (Withdrawn 2004). West Conshohocken, PA, 1988: ASTM International.
ASTM-E949. (2004). Standard Test Method for Total Moisture in a Refuse-Derived Fuel
Laboratory Sample (Withdrawn 2004). West Conshohocken, PA, 1988 ASTM International,.
ASTM-E1109. (2009). Standard Test Method for Determining the Bulk Density of Solid
Waste Fractions. West Conshohocken, PA, 1987: ASTM International.
Bernache-Pérez, G., Sánchez-Colón, S., Garmendia, A. M., Dávila-Villarreal, A., & Sánchez-
Salazar, M. E. (2001). Solid waste characterisation study in the Guadalajara Metropolitan Zone,
Mexico. Waste Management & Research, 19(5), 413-424. doi:10.1177/0734242x0101900506
Bolaane, B., & Ali, M. (2004). Sampling Household Waste at Source: Lessons Learnt in
Gaborone. Waste Management & Research, 22(3), 142-148. doi:10.1177/0734242x04044970
Brown, L. C., & Mac Berthouex, P. (2002). Statistics for environmental engineers: CRC press.
Byer, P. H., Hoang, C. P., Nguyen, T. T. T., Chopra, S., Maclaren, V., & Haight, M. (2006).
Household, hotel and market waste audits for composting in Vietnam and Laos. Waste
Management & Research, 24(5), 465-472. doi:10.1177/0734242x06068067
91
 Chan, W. W., & Lam, J. (2001). Environmental accounting of municipal solid waste
originating from rooms and restaurants in the Hong Kong hotel industry. Journal of Hospitality
& Tourism Research, 25(4), 371-385.
Chang, N.-B., & Pires, A. (2015). Technology Matrix for Solid Waste Management
Sustainable Solid Waste Management (pp. 19-97): John Wiley & Sons, Inc.
Dahlén, L., & Lagerkvist, A. (2008). Methods for household waste composition studies. Waste
Management, 28(7), 1100-1112. doi:http://dx.doi.org/10.1016/j.wasman.2007.08.014
Dangi, M. B., Pretz, C. R., Urynowicz, M. A., Gerow, K. G., & Reddy, J. M. (2011).
Municipal solid waste generation in Kathmandu, Nepal. Journal of Environmental Management,
92(1), 240-249. doi:http://dx.doi.org/10.1016/j.jenvman.2010.09.005
Dangi, M. B., Urynowicz, M. A., Gerow, K. G., & Thapa, R. B. (2008). Use of stratified
cluster sampling for efficient estimation of solid waste generation at household level. Waste
Management & Research, 26(6), 493-499. doi:10.1177/0734242x07085755
Denafas, G., Ruzgas, T., Martuzevičius, D., Shmarin, S., Hoffmann, M., Mykhaylenko, V., . . .
Ludwig, C. (2014). Seasonal variation of municipal solid waste generation and composition in
four East European cities. Resources, Conservation and Recycling, 89, 22-30.
doi:http://dx.doi.org/10.1016/j.resconrec.2014.06.001
E711-87, A. (2004). Standard Test Method for Gross Calorific Value of Refuse-Derived Fuel
by the Bomb Calorimeter (Withdrawn 2004). West Conshohocken, PA, 1987: ASTM
International.
Edjabou, M. E., Jensen, M. B., Götze, R., Pivnenko, K., Petersen, C., Scheutz, C., & Astrup, T.
F. (2015). Municipal solid waste composition: Sampling methodology, statistical analyses, and
case study evaluation. Waste Management, 36(0), 12-23.
doi:http://dx.doi.org/10.1016/j.wasman.2014.11.009
Espinosa Lloréns, M. d. C., Torres, M. L., Álvarez, H., Arrechea, A. P., García, J. A., Aguirre,
S. D., & Fernández, A. (2008). Characterization of municipal solid waste from the main landfills
of Havana city. Waste Management, 28(10), 2013-2021.
doi:http://dx.doi.org/10.1016/j.wasman.2007.07.004
Ezeah, C., Fazakerley, J., & Byrne, T. (2015). Tourism Waste Management in the European
Union: Lessons Learned from Four Popular EU Tourist Destinations. American Journal of
Climate Change, 4(05), 431.
Gallardo, A., Carlos, M., Peris, M., & Colomer, F. J. (2014). Methodology to design a
municipal solid waste generation and composition map: A case study. Waste Management,
34(11), 1920-1931. doi:http://dx.doi.org/10.1016/j.wasman.2014.05.014
Gidarakos, E., Havas, G., & Ntzamilis, P. (2006). Municipal solid waste composition
determination supporting the integrated solid waste management system in the island of Crete.
Waste Management, 26(6), 668-679. doi:http://dx.doi.org/10.1016/j.wasman.2005.07.018
Gomez, G., Meneses, M., Ballinas, L., & Castells, F. (2008). Characterization of urban solid
waste in Chihuahua, Mexico. Waste Management, 28(12), 2465-2471.
doi:http://dx.doi.org/10.1016/j.wasman.2007.10.023
92
 HASD. (2013). Statistical Yearbook - Hoi An city. Hoi An Statistical Department, Viet Nam.
Kathirvale, S., Muhd Yunus, M. N., Sopian, K., & Samsuddin, A. H. (2004). Energy potential
from municipal solid waste in Malaysia. Renewable Energy, 29(4), 559-567.
doi:http://dx.doi.org/10.1016/j.renene.2003.09.003
Longnecker, M., & Ott, R. (2001). An introduction to statistical methods and data analysis.
ISBN-13, 854576151.
MONRE. (2011). Báo cáo môi trường quốc gia 2011: Chất thải rắn (National Environment
Rerport 2011: Solid Waste). Hà Nội, Việt Nam: Bộ Tài nguyên và Môi trường.
Nguyen, T. K. T. (2014). Municipal solid waste management in Vietnam challenges and
solutions Municipal Solid Waste Management in Asia and the Pacific Islands (pp. 355-377):
Springer.
Nordtest. (1995). Solid waste, municipal: Sampling and characterisation (NT ENVIR 001).
Nordtest method (Publication no.
http://www.nordtest.info/index.php/methods/environment/item/solid-waste-municipal-sampling-
and-characterisation-nt-envir-001.html?category_id=24). http://www.nordtest.info
Philippe, F., & Culot, M. (2009). Household solid waste generation and characteristics in Cape
Haitian city, Republic of Haiti. Resources, Conservation and Recycling, 54(2), 73-78.
doi:http://dx.doi.org/10.1016/j.resconrec.2009.06.009
Qu, X.-y., Li, Z.-s., Xie, X.-y., Sui, Y.-m., Yang, L., & Chen, Y. (2009). Survey of
composition and generation rate of household wastes in Beijing, China. Waste Management,
29(10), 2618-2624. doi:http://dx.doi.org/10.1016/j.wasman.2009.05.014
Scheaffer, R., Mendenhall III, W., Ott, R., & Gerow, K. (2011). Elementary survey sampling:
Cengage Learning.
Shamshiry, E., Nadi, B., Bin Mokhtar, M., Komoo, I., Saadiah Hashim, H., & Yahaya, N.
(2011). Integrated models for solid waste management in tourism Regions: Langkawi Island,
Malaysia. J Environ Public Health, 2011.
Tchobanoglous, G., Theisen, H., & Vigil, S. (1993). Integrated solid waste management:
engineering principles and management issues: McGraw-Hill, Inc.
Teh, L., & Cabanban, A. S. (2007). Planning for sustainable tourism in southern Pulau
Banggi: an assessment of biophysical conditions and their implications for future tourism
development. Journal of Environmental Management, 85(4), 999-1008.
Thanh, N. P., Matsui, Y., & Fujiwara, T. (2010). Household solid waste generation and
characteristic in a Mekong Delta city, Vietnam. Journal of Environmental Management, 91(11),
2307-2321. doi:http://dx.doi.org/10.1016/j.jenvman.2010.06.016
Tran, D. T., Le, T. M., & Nguyen, V. T. (2014). Composition and generation rate of household
solid waste: Reuse and recycling ability. International Journal, 4(6), 73-81.
Trung, D. N., & Kumar, S. (2005). Resource use and waste management in Vietnam hotel
industry. Journal of Cleaner Production, 13(2), 109-116.
doi:http://dx.doi.org/10.1016/j.jclepro.2003.12.014
93
 Tuấn, N. V. (2014). Phân tích dữ liệu với R. T.P Hồ Chí Minh: Nhà xuất bản Tổng Hợp Thành
phố Hồ Chí Minh.
Van der Broek, E., & Kirov, N. (1971). The characterization of municipal solid wastes. Paper
presented at the Proc. Australian Waste Disposal Conference Sydney.
WB. (1999). Municipal Solid Waste Incineration. Retrieved from Washington, D.C:

94
5. MODELLING WASTE GENERATION PROGNOSIS MODEL

5.1 Introduction

An understanding of relationships between the quantity and quality of waste outputs from
human processes and local characteristics is essential in municipal solid waste management
(MSWM). To develop an adequate planning of sustainable waste management system, the
appropriate forecast of waste generation plays an important role. One of the challenges faced by
local governments is the prognosis of solid waste quantities to have appropriate actions and plan
(Ghinea et al., 2016). In Viet Nam, The National Technical Regulation QCVN 07:2010/BXD
(MOC, 2010) provided a method to estimate waste generation for five urban types based on
population and waste generation rate determined in the document. However, the results of
prediction are not reliable in term of practical application for different cities because the solid
waste generation is impacted not only by demographic factor but also by social, economic as
well as other factors (e.g. family expense or waste prevention policies). Therefore, the later
edition of this regulation, QCVN 07:2016 (MOC, 2016) uses neither this method for waste
generation estimation nor any other model instead. Lack of research and method on waste
generation estimation has led to a considerable challenge in municipal waste management in Viet
Nam.

Various modelling techniques were applied to develop solid waste generation, predictive
models. According to Beigl et al. (2008) and linear regression analysis, on the other hand, was
more popular in the application of waste generation estimation (Bach et al., 2004; Beigl et al.,
2008; Buenrostro et al., 2001; Ghinea et al., 2016; Thanh et al., 2010) due to the capacity of
expendability and modification to handle complexity of the real world, and it’s more useful in
empirical investigations and data prediction J. J. Faraway (2005).

Previous studies had modelled the total municipal waste generation using various variables
and reported that municipal waste generation rate significantly correlated with economic factors
(Hockett et al., 1995; Thøgersen, 1996), or household size (Abu Qdais et al., 1997; Dennison et
al., 1996), types of house (S Lebersorger et al., 2003) and the number of rooms in house
(Monavari et al., 2012). However, the diagnostic checking to verify the statistical adequacy of

95
the model was not carried out (Bdour et al., 2007; Karpušenkaitė et al., 2016; Shamshiry et al.,
2014) and model validation with a new dataset also got lack of attention (Buenrostro et al., 2001;
Ghinea et al., 2016; S. Lebersorger et al., 2011). The two analysis above was essential to
evaluate the reliability and performance of a linear regression model. In addition, Benítez et al.
(2008) chose the model based on the maximum R-square but in reality it is not a feasible option
when dealing with a lot of independent variables.

This study aims at to provide a reliable model to help the decision makers and related
stakeholders in forecasting the quantity of waste from households. The Bayesian Model Average
(BMA) method will be used instead of stepwise regression to avoid the noise variables from
gaining entry to the model in selecting predictor variables (Derksen et al., 1992). Multiple linear
regression models are developed with significant determinant variables chosen by BMA.
Diagnostic tests for the hypothesis of linear assumptions and a conventional validation method
were conducted to evaluate the performance of the model. The software R (v3.2.0) was used for
linear regression modelling and statistical analysis in this study (Julian J Faraway, 2005; J. J.
Faraway, 2005; Tuấn, 2014).

5.2 Mathematical modelling of Waste generation

The general procedure of the methodology applied is briefly described as follows. Firstly, a
household waste sampling and a questionnaire survey were carried out to gather the data of
waste generation and explanatory variables. Next, the BMA method was used to select the
significant predictor variables for the regression model. Then, the regression coefficients of the
model would be identified. Finally, the test for linear hypothetical assumptions and validation
was performed to evaluate the model’s performance.

5.2.1 Data collection

The waste generated was assessed from families in the whole town through a door-to-door
sampling method in 2015. To reduce variations in household waste generation, stratified random
sampling method was applied. The city was divided into two strata based on the Rural-Urban
topology of households. The number of samples was estimated from the number of households
in each stratum with a sampling ratio of 13 households per 1000 households (Hoang et al., 2017).
The needed statistical samples size should be 280 families; 321 households were gathered to
96
participate in the sampling program. Waste generated from households was sampled in 14 days
consecutively by 25 students of Da Nang University and two authors. Every household taking
part in the program was assigned with a code marker in front of the door to avoid collection
mistakes. The procedures for sample collecting and analysing are presented in Figure 4.3.

Then, a survey was conducted by face-to-face interview questionnaire to all households in

the project. The study is to obtain information of personal and socioeconomic background of the
family such as house area, with garden or not, family size, member ages, and monthly income,
etc. The results obtained from this survey were interpreted to identify the explanatory variables
for the mathematical models of household waste generation estimation.

5.2.2 Variables used in modelling

Regression analysis was used for explaining the relationship between dependent variable Y
(or response or output variable), and one or more independent variables (or predictor variables)
X. Thus, analysis of data gathered from the questionnaire survey was followed by identifying the
variables involved. The database comprised ten independent variables presenting demographic,
geographic and economic influencing factors to household waste generation. The response
variable is mean of the per capita waste generation (kg capita-1 day-1) gathered from the sampling
of household waste. Three types of independent variables including categories, discrete and
continuous variables were employed. Table 5.1 explains the variables included and symbols
assigned. A variable matrix consisting the information of household-family participating in the
sampling was constructed and used as the input dataset for the analysis.

97
 Table 5.1 Types of variables in linear regression

Type of variable Variable Symbol Unit/value/types Determination of variables

response variable
Dependent Per capita waste Yhhw kg per capita per Average of daily waste generated from
continuous generation per day each family member in each
day household
Predictor variables

Independent category Household Xplc Urban (=1) If the house located in urban area
location Countryside (=0) If the house located in country area
Independent category House garden Xgar Yes (=1) The house has garden
No (=0) The house does not have garden
Independent category Home business Xbus Yes (=1) Family members do business at home
such as a convenience store,
restaurant, café bar, shop, mini hotel,
vehicle rent
No (=0) Do not have any business at home
Independent discrete Family income Xinc 1 Very low: less than 500.000 VND per
person per month
2 Low: From 500.000 - 1.200.000 VND
per person per month
3 Lower-middle: From 1.200.000 –
2.500.000 VND per person per month
4 Upper-middle: From 2.500.000-
4.000.000 VND per person per month
5 High: From 4.000.000- 6.000.000
VND per person per month
6 Very high: more than 6.000.000 VND
per person per month
Independent discrete Household size Xsiz Number The number of individuals in the
family
Independent discrete Number of room Xrom Number Number of rooms in the house
Independent House area Xare m2 The total area of the house
continuous
Independent House area per Xpa m2 per person The household area divided by the
continuous person number of family members
Independent % of children Xchi Percentage The percentage of people less than 20
continuous years old in the family
Independent % of adults Xadu Percentage The percentage of people from 20 - 59
continuous years old in the family
Independent % of old people Xold Percentage The percentage of people older than 59
continuous years old in the family

5.2.3 Selection of determinant variables

The dataset was divided randomly into two sets: 70% of the data as training set and other
30% is testing set. The training sample set would be used to determine the predictor variables
and identify the coefficient of the model. Meanwhile, the testing sample set will be used for

98
validating the model. The Bayes model average method (Raftery et al., 1997, Hoeting et al.,
1998) will be utilised to identify the combination of significant independent variables to explain
MSW generation best. The “best” model can provide the most precise prediction with a
reasonable number of variables or accurate estimations for new cases exchangeable (Raftery et
al., 1997). BMA provides a consistent mechanism of accounting for this model uncertainty
which is often ignored in model selection, leading to overfitting models and might cause over-
confident inferences (Hoeting et al., 1999; Fernández et al., 2001). According to Hoeting et al.
(1999), BMA also provides improved the out-of-sample predictive performance of the linear
model. A Bayesian Model Averaging solution to this problem provides optimal predictive ability
by involving averaging over all possible models (Madigan and Raftery, 1994). Quantities of
interest and parameters estimates are provided via direct application of the principles described
as follows:

The posterior distribution given data Z of the quantity of interest  , such as a model
parameter or a future observable is defined by Eq.5.1.
K
p(  | Z )   pr (  | M k , Z ) pr ( M k | Z )
k 1 (5.1)

Where, M1, M2, ..., MK are the models under consideration. Eqs.5.2 and 5.3 give the posterior
probability for model Mk and the integrated likelihood of Mk

pr ( Z | M k ) pr ( M k )
pr ( M k | Z )  K (5.2)
 pr( Z | M
k 1
k ) pr ( M k )

pr ( Z | M k )   pr ( Z |  k , M k ) pr ( k | M k )d k (5.3)

Where,  k is the vector of parameters of model Mk and pr( k | M k ) is the prior density of the

parameters under the model, pr ( D |  k , M k ) is the likelihood and pr ( M k ) is the prior probability
that Mk is the actual model. The BMA estimates a parameter  by Eq. 5.4.

^ K
 BMA    k pr( M k | Z ) (5.4)
k 1

99
 The Bayesian information criterion called BIC approximation which is formally defined as
Eq.5.5. BIC is used as a criterion for models selection from the set of model, the model with the
lowest BIC approximation is preferred.

BIC = -2. log (RSSp) + p.log n (5.5)

Where,

BIC: is Bayesian information criterion

RSSp: is square sum of residuals in fitting sample data to the model with p independent
variables

p: is the number of regressors including the intercept

n: is the number of observation or equivalently, the sample size

5.2.4 Multivariate linear regression model

A multivariate linear regression model is described as Eq. 5.6.

Y = α + β1X1 +β2X2 +β3X3 +… + ε (5.6)

Where, α is the intercept term which indicates the mean of the dependent variable Y in the
case that all predictor variable X is equal to 0; β, a vector of β i, is the slope of the model, and it
explains the average change in the dependent variable. The residual ε represents the difference
between estimated values and observed values. ε may include measurement error although it is
often due to the effect of unincluded or unmeasured variables (Faraway, 2005). In linear
regression, it is assumed that the errors are normally distributed, independent and have equal
variance σ2 (ε~N(0,σ2I)). The correlation of residuals is vital for time series data because time
series regression accounts for the autocorrelation between time. Meanwhile, in non-time series
regression, the independence of errors is presumed or at least minimised. Theoretically, the
residuals from the model should not be correlated with either independent or dependent variables.

100
5.2.5 Testing model assumptions

The validity of the assumptions underlying the chosen model should be checked. The
residuals ε was used to test the linear model assumptions. Formal diagnostic tests can ensure the
exactitude of results but may be powerless to detect problems of an unsuspected nature,
especially in data related to social and human activities. Graphical techniques are usually more
efficient at revealing overall structure of the data set. They tend to be more versatile and
informative (Faraway, 2005). Moreover, the graphical methods may be useful for describing as
well as understanding the underlying structure of the data (Wilk and Gnanadesikan, 1968).
Therefore, in this study, the graphical approach was applied as diagnostic tests for the hypothesis
of linear model assumptions. The normality of residual distribution is tested by normal quantile
plot of the residuals (Wang and Bushman, 1998). The plot consists of plotting the ordered
residuals from the fitted model (vertical axis) versus the reference line of a normal distribution
having the same mean and variance (horizontal axis). The points of model residuals on such a
plot should fall close to the reference line if the errors are normally distributed. Violation of
normality often occurs either because the distributions of either predictor or response variable are
significantly not normal. In this article, author generated the residuals versus fitted values plot
and residuals versus independent variables plot to find out the way to improve the model. A
useful method is that if the non-random shape occurs in only one of the plots, the predictor
variable should be transformed. If it happens in more than one plots, we should transform the
response variable to improve the model. “The plot of residuals versus estimated values (fitted
values)” can also indicate constant variance if the scatter is symmetric vertically around zero.
There is some approach to dealing with non-constant variance violation in linear regression
model. Weighted least squares or transformation of the response variable can be used to have a
constant variance of the outcome variable (Faraway, 2005). Likewise, to test for violations of
independence, the residuals distribution should be random and symmetric around zero under all
conditions. Outlier observations which do not fit the model and influential observations that have
large effects on the model will be detected. The outlier test was carried out by Bonferroni

101
correction method (Faraway, 2005) and the Cook statistic was used for influence diagnostic
(Cook, 1977).

5.2.6 Model evaluation and validation

A conventional validation approach with external validation method will be applied to test
the model for avoiding over-fitting (Faber and Rajkó, 2007). The requirement is that the
validation samples are entirely different from the training samples which constructed the model.
Thus, this approach properly necessary to assess the ability to forecast for unknown future
samples (Bleeker et al., 2003; Faber and Rajkó, 2007). To ensure the model can perform in a
new dataset, authors divided the original data into two subsets including training set (70%) and
testing set (30%). The former is used for constructing the model, whereas the latter is for
validating. A combination of statistical metrics would be applied to assess the model
performance, which is including coefficient of determination R 2, adjusted R2 (R2adj), mean
absolute error (MAE), root means square error (RMSE) and normalised root mean square error.
R2 (R-squared) is a useful property indicating the goodness of fit of the model. The R2adj
(adjusted R-squared), on the other hand, also indicates how well model fit, but adjusts for the
number of independent variables in a model. MAE and RMSE are useful measures widely used
in model evaluations. MAE considers the same weight to all kind of errors which is proper to
describe uniformly distributed errors. Meanwhile, RMSE favours the errors with larger absolute
value and is appropriate to explain normally distributed error (Chai and Draxler, 2014). In this
study, RMSE was used for assessing the performance of predictive model since the residuals of
the linear regression model is expected to be normally distributed. If the RMSE for the testing
data set is higher significantly than that of the training data set, over-fitting occurs. The two
RMSEs are close, which mean that the model is valid and enable to predict unknown data.
However, the range of training data is different from testing data. Thus, to compare the RMSE of
two data set, the normalised RMSE (NRMSE) was used. NRMSE is the ratio of the RMSE to the
range of the data set, so NRMSE has the range of 0 to 1. Eqs. 7 and 8 explain RMSE and
NRMSE, respectively. MAE will be calculated to describe the average error magnitude.

102
 n

 ( yˆ i  yi ) 2
RMSE  i 1
(7)
n

RMSE
NRMSE  (8)
y max  y min

Where, ŷi : the estimated value of the outcome variable of observation i

yi : the observed value of the dependent variable observation i

y max : the maximum observed value of the dependent variable

y min : the minimum observed value of the dependent variable

n: the sample size.

A bootstrapping with 10.000 replications on the training data set is carried out to calculate
the 95% confidence interval of R-square (95% ci). The coefficient of determination of the model
running on testing data is also calculated. R-square of the model on testing data is expected to
belong to the 95% ci.

5.3 Result and discussion

5.3.1 Significant independent variables and selected models

The number of households responded to the face-to-face interview questionnaire survey was
286 over 321, which was more than the statistical requirement for sample sizes (281). Thus, the
nonresponse samples were just neglected in later analysis. Figure 5.1 shows the results of
estimation of correlation coefficients among variables. The results indicate that the correlation
coefficients of outcome variables and explanatory variables are low (<0.33). In Figure 5.2, the
horizontal axis names options chosen by BMA. The red colour indicated that the predictors
variable correlated with the outcome variable with a positive coefficient. The blue colour
presented the negative correlation, and the other colour means that the variable is not present in
the model. The result of BMA method indicated that there are four independent variables
including Household location (Xplc), Home business (Xbus), Household size (Xsiz) and house area

103
per person (Xpa), proved to be significant variables for estimating daily waste generation per
capita (Figure 5.2). Two variables such as Xplc and Xsiz were presented in all group of
significantly determinant variables selected by BMA (p = 100) while the probability of the
variable Xbus appearing in models chosen by BMA is 96.4% and that of X pa is about 74%.
Interestingly, the variables presented house income, garden, and percentages of members of
different ages range in the family were not significant, indicating that those factors do not
explain the variation in the waste generation rate.

Figure 5.1 Correlation coefficients among variables

The household size (Xsiz) negatively correlated with daily per capita waste generation. An
increase of family members will lead to a decrease of daily waste generation per capita. Previous
studies were consistent with our finding of the qualitative relationship between household size

104
and per capita daily waste generation per capita. Benítez et al. (2008); Qu et al. (2009) and
(Sukholthaman et al., 2015) found the same negative influence of household size on waste
generation rate per capita. The positive correlation between the regressor (Xplc) and the response
variable indicated that the household waste generation rate associated with the area it is located.
People living in urban area generated more waste than ones living in countryside area. In contrast,
Hockett et al. (1995) found that the type of urbanisation was not a significant determinant of
waste generation rate in his study. In Vietnam, the home served as a base for a business such as a
convenience store, a restaurant, or a place for manufacturing goods, is quite familiar. The
presence of business at home might affect to waste generation per capita quantity (Parizeau et al.,
2006). Home business (Xbus) was confirmed to be the significant determinant of household waste
generation rate (Table 2). The variable X bus correlated positively with waste generation means
that family handling business at home has higher daily waste generation per person rate than one
without business. Higher income leads to more goods consumption and therefore more waste
production (Buenrostro et al., 2001). Nevertheless, other researchers found that the household
income is not related to waste generation by different measurement of income approaches such
as continuous income data (Bernache-Pérez et al., 2001; Benítez et al., 2008; Grazhdani, 2016),
categorical income data (Bolaane and Ali, 2004, Gomez et al., 2008), or proxy variables
(Mbande, 2003; Gomez et al., 2008; Prades et al., 2014). The result of this study indicated that
direct income variable (Xinc) is not a significant determinant factor of waste generation. The
difficulty in this relationship investigation is the accuracy of income data solicited from
households, especially in developing countries (Parizeau et al., 2006). In HAC, people might
consider their income to be a private matter, and business households try to conceal the real
income to avoid paying more taxes. A proxy variable of the income, the total area of the house
(Xare) was not a significant variable to estimate the per capita waste generation. Meanwhile, the
house area per capita (Xpa) was proved to be a determinant variable of household waste
generation rate. It meant that the amount of waste produced is correlative to the average house
space for a person. The number of the room in the house was not an explanatory variable for
waste generation estimation as a result of BMA. It is inconsistent with the previous study, the
production of household waste was found to be positively correlated with the number of the
room (Monavari et al., 2012). Other variables representing home garden and age groups were not

105
significantly correlated with the quantity of waste produced. The Bayesian Model Averaging
method not only detected the best model but also suggested some other reliable models for
predicting household waste generation based on BIC approximation. Thus, we could have some
model options to identify the amount of waste. Table 2 showed the best five options of waste
generation prognosis models suggested by BMA. The model 1 with four predictors has the
lowest BIC approximation (-62.3), which means that the linear regression model of four
variables (Xplc, Xbus, Xsiz, and Xpa) is the best multivariate model among all possibilities.

Figure 5.2 Predictors being chosen for the most reliable models by BMA

Posterior probability presents the likelihood of the model that may explain the observed data
correctly. The posterior probability of models 1 and 2 is higher than other models, approximately
42% (0.422) and 19% (0.187), respectively. It indicated that models 1 and 2 are more accurate in
explanation of waste generation observations than other models (around 5%). The model with
three independent variables has lower R 2adj (of about 30%), and models of four and five
regressors have adjusted R-square of about 33%, which are similar. It means that adding more
than four independent variables into the model will not improve the model fit. Table 5.2 also
showed that significant level of each variable in each single model. Each of model 3,4 and 5 has
more than four regressors, but not all the explanatory variables are significant determinants. The
variable Xrom (Model 3), Xgar and Xare (Model 4 and 5) were proven negligible determinants for
waste generation. Thus, we choose Model 1 and 2, whose all predictor variables are significant,

106
lower BIC approximations and higher posterior probability to be tested for the linear model
assumption and predictive performance.

Table 5.2 Best models selected by Bayesian Model Average

Variables Model 1 Model 2 Model 3 Model 4 Model 5

Intercept -1.539 (***) -1.293 (***) -1.402 (***) -1.894 (***) -1.517 (***)
Xplc(Urban) 0.578 (***) 0.535 (***) 0.560 (***) 0.457 (***) 0.557 (***)
Xare - - - 0.0005( ) -
Xrom - - -0.055 ( ) - -
Xgar(YES) - - - -0.08 ( ) -0.061( )
Xsiz -0.128 (***) -0.147 (***) -0.120 (***) - -0.125 (***)
Xpa 0.004 (**) - 0.004 (**) - 0.004 (**)
Xinc - - - - -
Xbus(YES) 0.302 (**) 0.317 (***) 0.308 (**) 0.292 (**) 0.304 (**)
Number of 4 3 5 4 5
variables used
BIC -62.3 -60.7 -58.8 -57.8 -57.5
Posterior 0.422 0.187 0.071 0.043 0.038
probability
R2 0.343 0.319 0.348 0.328 0.344
R2adj 0.329 0.309 0.332 0.318 0.327
F-statistic 25.43 30.64 20.76 10.61 20.38
Signif. Codes: p ~ 0 (***); p < 0.001 (**); p < 0.05 (*); p < 0.1 ( . ); p < 1 ( )

5.3.2 Multivariate linear regression models for household waste generation

Eqs.5.8 and 5.9 show the parameter estimates for the selected model 1 and model 2,
respectively:

Log (YHHW) = -1.539 + 0.578Xplc(Urban) – 0.128Xsiz + 0.004Xpa + 0.302Xbus(YES) (5.8)

Log (YHHW) = -1.293 + 0.535Xplc(Urban) – 0.147Xsiz + 0.317Xbus(YES) (5.9)

The intercept -1.539 (in the equation 8) is the unconditional expected mean of the logarithm
of waste generation rate. Therefore the exponential value of the intercept equals to 0.215
(kg/capita/day) is the geometric mean of waste generation rate. The exponential value of the
coefficient for Xplc is 1.78 (e0.578 = 1.78), it indicates that average of the per capita waste
generation of household in an urban area is 78% higher than that of household in the rural area
when other independent variables are held at some fixed values. Similarly, a person in a family
doing business at home generated 35% more than the one living in a family not doing business
107
(exponential value of 0.302 equals to 1.35). The household size is the only significant predictor
variable negatively correlated with waste produced, indicating that an increase in the number of
family members is associated with a decrease in waste generation per capita. The coefficient of
variable Xsiz in the model is -0.128, it means that one person increase in the family leads to a
decrease of 12% in waste generation rate (e-0.128 = 0.88) when other variables values are
unchanged. Parameter explanation is similar to model 2. Figure 5.3 describes the predicted value
of waste generation per capita from the Model 1 with four predictor variables. Different lines in
the plot present the estimated waste generation rate for the various values of variable household
size. In this figure, we showed four value of X siz (Xsiz = 3,4,5 and 6). Moreover, Figure 5.4
describes the predicted values of waste generation rate based on the three independent variable
model chosen.

The purpose of this study is to find out a simple and reliable model to estimate waste
generation, which can contribute to improving waste management. The models can provide
reliable information to support current waste collection and transportation. Exact estimation of
waste generation in rural and urban areas results in better design and arrangement of vehicles,
labours, collection routes. That will improve the efficiency of the current collection system,
which has been inefficient due to poor calculation and design. Moreover, as a result of model
estimation, a decentralised management approach could benefit the waste collection due to the
variation of waste generation amount in urban and countryside areas. Onsite-treatment or small
scale treatment might reduce the cost of collecting the little amount of waste generated in further
areas, especially degradable waste have a great potential for home-composting or recycling for
feeding animals in agriculture places.

The result of the study also suggested that home businesses contributed considerable
amounts of waste to the total waste requiring being collected from a household. Therefore,
estimation of waste generation from houses with business could provide a basis to the decision-
makers to improve the waste management system such as giving more weight to commercial and
tourism sectors in the city. For instance, increasing waste collection fee for home and business
sectors can support waste management since the number of households involved in this business
activities are increasing quickly.

108
Figure 5.3 Fitted line plot of model with four regressors Xplc, Xsiz, Xpa, Xbus

109
 Figure 5.4 Fitted line plot of model with three regressors Xplc , Xsiz, Xbus

5.3.3 Models analysis

5.3.3.1 Diagnostic test for linear model assumption

Figure 5.5 represents the test results for normality, constant variance, and autocorrelation.
Two quantile-quantile plots (Figures 5.5.1 and 5.5.7) compare the residuals (all the scatters on
the graph) to “ideal” normal observations (the line). The residuals follow the line approximately
indicating that the errors of both models are normal. The residual versus fitted value plot
(Figures 5.5.2 and 5.5.8) used for a test of nonconstant variance. The scatter is symmetric
vertically around zero demonstrating that there is no evidence of nonconstant variance. Moreover,
Figures 5.5.2, 5.5.3, 5.5.4, 5.5.5 and 5.5.6 (model 1) and Figures. 5.5.9, 5.5.10 and 5.5.11 (model
2) show that the scatter of residuals are symmetric approximately around zero in the plot with all
the independent variables. It means that there is no problem with the correlation of the residuals.
The results from graphical test indicated that all linear assumptions were satisfied.

5.3.3.2 Influences of variables and observations

Figure 5.6 shows the relative importances (with 95% confidence interval) of regressors for
the two models, performed by LMG method (Johnson and LeBreton; 2004, Grömping, 2006).
The place of household (Xplc) acts as the primary predictor variable in both models because the
percentage of contribution is about 40%, followed by the household size (X siz), around 30%. The

110
independent variables have less contribution to waste generation (10% and 20%) was household
area per person (Xpa) and home business (Xbus), respectively.

Figure 5.5 Tests for linear assumption of two models

111
 Figure 5.6 The relative importance of regressors for waste generation rate

Figure 5.7 indicated the observations have a large impact on the predicted values, measured
by Cook’s distance (Cook, 1977, Cook, 1979). The observations number 13, 70 and 184 have
significant influences to the fit of the models than other observation, but none of them has too
much influence (Cook’s distance less than 1.0), (David, 2007). Outlier tests showed that there
was one outlier observation in both models, which is the observation number 47.

Figure 5.7 Observations have high influence to the model

112
5.3.3.3 Model validation

The R2 of the models were 0.34 (model 1) and 0.32 (model 2), it means that the models
could explain about 34% and 32% the variation of daily waste generation rate per capita,
respectively. Other multivariate linear regression studies had low R 2 such as 51% in the study of
Benítez et al. (2008), 36% in the research of Grossman et al. (1974) and 48.7% in the study by
Bach et al. (2004). The weak coefficient of determination could be explained by the fields of
study, especially in waste generation study, which attempts to predict human behaviour such as
habit, lifestyle, normally has R 2 values lower than 50%. Humans behaviour are simply harder to
predict than physical processes. However, the goal is not to maximise the coefficient of
determination because obtaining more predictor variables may cause overfitting. In other words,
a low R-square does not mean the model is useless, and a significant R-square is not able to say
the model is useful (Brown and Mac Berthouex, 2002). Also, a good model can maximise the
percentage of variation explained but limits the ability of the results to be generalised (Beigl et
al., 2008). If the model satisfies all the linear regression assumptions, it is a proper model to
estimate waste generation. Moreover, it can draw some meaningful conclusions about how
changes in the predictor variables are associated with variations in the response variables.

Table 5.3 Results of model validation

Models Datasets Standard deviation R2 Mean R2 (bootstrap) 95% ci of R2 RMSE NRMSE MAE
Model 1 Train set 0.712 0.343 0.354 0.235 – 0.428 0.576 0.131 0.451
Test set 0.281 0.678 0.137 0.488
Model 2 Train set 0.712 0.319 0.327 0.220 – 0.406 0.586 0.134 0.453
Test set 0.256 0.689 0.139 0.492
Multivariate linear regression model created by training data set was run on testing data set,
the statistical metrics of performance was calculated. Table 5.3 explained the model validation
results. In both models, R-square of testing data set are in the 95% ci, NRMSEs and MAE of
training and testing set are very close, which means two models perform well on new data. On
the other hand, RMSEs are smaller than the standard deviation of the response variable (0.712),
indicates that the model produces less variation than the data observations. Lastly, low NRMSEs
(about 0.13 over the range of 0 to 1) demonstrate that the fitted values of both models are quite
close to the observations. Thus, both models have good performance in predicting the household

113
per capita waste generation. Also, model 1 has performed better than model 2 because it has
higher R-square, higher posterior probability and produces smaller errors (RMSE, MAE).

5.4 Conclusion

The models constructed in the current study is valuable in estimating waste generation as it
provides observational evidence of the influences of multiple factors. They indicated that the
impacts of socio-demographic, geographic variables and family economic activities on waste
generation rate are highly significant. The model can not predict waste generation in the future,
but it can provide reliable information to improve the current waste management. The location of
the house is the most affected predictor for the daily amount of waste generated per capita. Both
models showed that a person in an urban house produced much higher solid waste (70-80%) than
one living in a countryside family. That provides an exact estimation of waste generation in rural
and urban areas in better calculation and arrangement of vehicles, labours, collection routes. It
also suggests that a decentralised treatment approach could benefit to reduce the collection cost
for the small amount of waste generation in further areas. However, the approach needs to be
studied carefully to have the applicable implications and appropriate legislation to encourage
decentralised waste treatment. Education and awareness of waste generation were implemented
successfully in Hoi An and should be maintained and improved. Another important factor
influenced the household waste generation in HAC is family economic activities. The result of
the two models showed that if a family is holding a business at home, the waste generation rate
of the household will increase by about 35%. Waste fee for business sector might be an
important factor to take into consideration in waste collection and management planning when
the number of household opening business at home for tourism and local services such as small
restaurants, homestays, shops, convenience stores and vehicle rental shops have been rising
gradually in the city. The study found that household size and the area of the house are also
significant determinants of per capita waste generation while other variables, particularly income
per person, prove to not significant as correlates of waste production. Results from this study
demonstrate that Bayesian Model Average method was a robust method to determine solid
options in multiple linear regression models, especially dealing with a big number of
independent variables. The result of BMA indicated that the linear regression model with four

114
independent variables (X plc, Xsiz, Xpa, Xbus) was the best model for estimating waste generation in
HAC because of the lowest BIC approximation and reasonable model size. Meanwhile, the
model with three regressors (Xplc, Xsiz, Xbus) has slightly lower performance but is very useful to
predict the waste generation of household quickly because of availability information of
predictor variables in census database.The study attempted to increase the understanding of
waste generation to support waste management planning for the city. Thus, the two models are
useful in addressing waste generation for not only analysing key factors influencing the waste
generation, but they also giving waste managers a link to have a model for estimating waste
generation to improve waste reduction and management efforts. The results and methodology are
accepted to be informative for authorities, decision makers, stakeholders and planners to develop
a waste management plan. The model developed in this paper is not reliable for predicting waste
generation in the future. Lack of historical data in caused difficulties for modelling waste
predictive model. Thus, modelling municipal waste generation model which can forecast the
future is essential and needed to be studied.

5.5 Reference

Abbasi, M.; Abduli, M.; Omidvar, B.; Baghvand, A., (2012) Forecasting municipal solid
waste generation by hybrid support vector machine and partial least square model. Int. J. Environ.
Res., 7(1): 27-38 (12 pages).
https://journals.ut.ac.ir/article_583_2e7226656aec9ea041dc6023993c8d64.pdf
Abdoli, M.A.; Rezaei, M.; Hasanian, H., (2016). Integrated solid waste management in
megacities. Global J. Environ. Sci. Manage., 2(3): 289-298 (10 pages).
http://www.gjesm.net/article_18857_8ff02e369f0d786a518d2a5fffdec380.pdf
Abu Qdais, H.A.; Hamoda, M.F.; Newham, J., (1997). Analysis of Residential Solid Waste At
Generation Sites. Waste. Manage. Res., 15(4): 395-406 (11 pages).
http://journals.sagepub.com/doi/abs/10.1177/0734242X9701500407
Akhtar, S.; Saleem, W.; Nadeem, V.M.; Shahid, I.; Ikram, A., (2017). Assessment of
willingness to pay for improved air quality using contingent valuation method. Global J. Environ.
Sci. Manage., 3(3): 279-286 (17 pages).
http://www.gjesm.net/article_24692_feadd09a34a7cbd8872e4dfe46e33712.pdf
Bach, H.; Mild, A.; Natter, M.; Weber, A., (2004). Combining socio-demographic and logistic
factors to explain the generation and collection of waste paper. Resour. Conserv. Recyc., 41(1):
65–73 (9 pages).

115
 http://www.sciencedirect.com/science/article/pii/S0921344903001216
Bdour, A.; Altrabsheh, B.; Hadadin, N.; Al-Shareif, M., (2007). Assessment of medical wastes
management practice: a case study of the northern part of Jordan. Waste. Manage., 27(6): 746–
759 (14 pages).
http://www.sciencedirect.com/science/article/pii/S0956053X06000948
Beigl, P.; Lebersorger, S.; Salhofer, S., (2008). Modelling municipal solid waste generation: A
review. Waste Manage, 28(1): 200-214 (15 pages).
http://www.sciencedirect.com/science/article/pii/S0956053X07000153
Benítez, S.O.; Lozano-Olvera, G.; Morelos, R.A.; Vega, C.A.d., (2008). Mathematical
modeling to predict residential solid waste generation. Waste Manage., 28, Supplement 1: S7-
S13 (7 pages).
http://www.sciencedirect.com/science/article/pii/S0956053X08001438
Bernache-Pérez, G.; Sánchez-Colón, S.; Garmendia, A.M.; Dávila-Villarreal, A.; Sánchez-
Salazar, M.E., (2001). Solid waste characterisation study in the Guadalajara Metropolitan Zone,
Mexico. Waste Manage. Res., 19(5): 413– 424 (15 pages).
http://journals.sagepub.com/doi/abs/10.1177/0734242X0101900506
Bleeker, S.E.; Moll, H.A.; Steyerberg, E.W.; Donders, A.R.T.; Derksen-Lubsen, G.; Grobbee,
D.E.; Moons, K.G.M., (2003). External validation is necessary in prediction research:: A clinical
example. J. Clin Epidemiol., 56(9): 826-832 (7 pages).
http://www.sciencedirect.com/science/article/pii/S0895435603002075
Bolaane, B.; Ali, M., (2004). Sampling Household Waste at Source: Lessons Learnt in
Gaborone. Waste Manage. Res., 22(3): 142–148 (7 pages).
http://journals.sagepub.com/doi/abs/10.1177/0734242X04044970
Boulet, S.; Boudot, E.; Houel, N., (2016). Relationships between each part of the spinal curves
and upright posture using Multiple stepwise linear regression analysis. J. Biomec., 49(7): (7
pages).
http://www.sciencedirect.com/science/article/pii/S0021929016302512
Brown, L.C., Mac Berthouex, P., (2002). Statistics for environmental engineers. CRC press.
https://www.crcpress.com/Statistics-for-Environmental-Engineers-Second-Edition/Brown-
Berthouex/p/book/9781566705929
Buenrostro, O.; Bocco, G.; Vence, J., (2001). Forecasting Generation of Urban Solid Waste in
Developing Countries—A Case Study in Mexico. J. Air. Waste Manage. Assoc., 51(1): 86-93 (8
pages).
http://www.tandfonline.com/doi/abs/10.1080/10473289.2001.10464258
Chai, T.; Draxler, R.R., (2014). Root mean square error (RMSE) or mean absolute error
(MAE)?–Arguments against avoiding RMSE in the literature. Geosci. Model Dev., 7(3): 1247-
1250 (14 pages).
116
 https://www.geosci-model-dev.net/7/1247/2014/gmd-7-1247-2014.pdf
Chang, Y.F.; Lin, C.J.; Chyan, J.M.; Chen, I.M.; Chang, J.E., (2007). Multiple regression
models for the lower heating value of municipal solid waste in Taiwan. J. Environ. Manage.,
85(4): 891-899 (9 pages).
http://www.sciencedirect.com/science/article/pii/S0301479706003562
Chen, H.W.; Chang, N.-B., (2000). Prediction analysis of solid waste generation based on grey
fuzzy dynamic modeling. Resour. Conserv. Recyc., 29(1-2): 1-18 (18 pages).
http://www.sciencedirect.com/science/article/pii/S092134499900052X
Chu, M.T., (2014). Báo cáo tổng kết đề tài: Xây dựng mô hình đồng quản lý rác thải tại hai xã,
phường Cẩm Hà và Cẩm Phô, thành phố Hội An (Project report: building co-operative waste
management model in two wards, Cam Ha and Cam Pho, Hoi An city). Management board of
Cham islands MPA, Hoi An, Viet Nam.
http://hoian.gov.vn/CMSPages/BaiViet/Default.aspx?IDBaiViet=12905
Cook, R.D., (1977). Detection of Influential Observation in Linear Regression. Technometrics,
19(1): 15-18 (4 pages).
http://www.tandfonline.com/doi/abs/10.1080/00401706.1977.10489493
Cook, R.D., (1979). Influential Observations in Linear Regression. J. Am. Stat. Assoc.,
74(365): 169-174 (6 pages).
http://www.tandfonline.com/doi/abs/10.1080/01621459.1979.10481634
Daskalopoulos, E.; Badr, O.; Probert, S.D., (1998). Municipal solid waste: a prediction
methodology for the generation rate and composition in the European Union countries and the
United States of America. Res. Conserv Recyc., 24(2): 155-166 (12 pages).
http://www.sciencedirect.com/science/article/pii/S0921344998000329
David, M.L., (2007). Influential Observations, Online Statistics Education: A Multimedia
Course of Study, Rice University.
http://onlinestatbook.com
Dennison, G.J.; Dodd, V.A.; Whelan, B., (1996). A socio-economic based survey of
household waste characteristics in the city of Dublin, Ireland — II. Waste quantities. Resour.
Conserv. Recyc., 17(3): 245-257 (13 pages).
http://www.sciencedirect.com/science/article/pii/092134499601155X
Derksen, S.; Keselman, H.J., (1992). Backward, forward and stepwise automated subset
selection algorithms: Frequency of obtaining authentic and noise variables. Brit. J. Math. Stat.
Psychol., 45(2): 265–282 (18 pages).
http://onlinelibrary.wiley.com/doi/10.1111/j.2044-8317.1992.tb00992.x/full
Faber, N.M.; Rajkó, R., (2007). How to avoid over-fitting in multivariate calibration—The
conventional validation approach and an alternative. Anal. Chim. Acta., 595(1–2): 98-106 (9
pages).
117
 http://www.sciencedirect.com/science/article/pii/S0003267007009129
Fernández, C.; Ley, E.; Steel, M.F.J., (2001). Model uncertainty in cross-country growth
regressions. J. Appl. Econ., 16(5): 563-576 (14 pages).
http://onlinelibrary.wiley.com/doi/10.1002/jae.623/epdf
Ghinea, C.; Drăgoi, E.N.; Comăniţă, E.-D.; Gavrilescu, M.; Câmpean, T.; Curteanu, S.;
Gavrilescu, M., (2016). Forecasting municipal solid waste generation using prognostic tools and
regression analysis. J. Environ. Manag., 182: 80-93 (14 pages).
http://www.sciencedirect.com/science/article/pii/S0301479716304637
Gomez, G.; Meneses, M.; Ballinas, L.; Castells, F., (2008). Characterization of urban solid
waste in Chihuahua, Mexico. Waste Manage., 28(12): 2465-2471 (7 pages).
http://www.sciencedirect.com/science/article/pii/S0956053X07004126
Grömping, U., (2006). Relative importance for linear regression in R: the package relaimpo. J.
Stat. Software, 17(1): 1-27 (27 pages).
https://core.ac.uk/download/pdf/6305006.pdf
Grazhdani, D., (2016). Assessing the variables affecting on the rate of solid waste generation
and recycling: An empirical analysis in Prespa Park. Waste Manage., 48: 3-13 (11 pages).
http://www.sciencedirect.com/science/article/pii/S0956053X15301379
Grossman, D.; Hudson, J.; Marks, D., (1974). Waste generation models for solid waste
collection. J. San. Eng. Division, 100(6): 1219-1230 (12 pages).
http://cedb.asce.org/CEDBsearch/record.jsp?dockey=0021718
Hamby, D.M., (1994). A review of techniques for parameter sensitivity analysis of
environmental models. Environ. Monit. Assessment, 32(2): 135–154 (10 pages).
https://link.springer.com/article/10.1007/BF00547132
HASD, 2013. Statistical Yearbook - Hoi An city, Hoi An Statistical Department, Viet Nam.
http://hoian.gov.vn/CMSPages/BaiViet/Default.aspx?IDBaiViet=12888
Hoang, M.G.; Fujiwara, T.; Pham Phu, S.T., (2017). Municipal waste generation and
composition in a tourist city - hoi an, vietnam. J. SCE, 5(1): 123-132 (10 pages).
https://www.jstage.jst.go.jp/article/journalofjsce/5/1/5_123/_article
Hockett, D.; Lober, D.J.; Pilgrim, K., (1995). Determinants of Per Capita Municipal Solid
Waste Generation in the Southeastern United States. J. Environ. Manage., 45(3): 205-217 (13
pages).
http://www.sciencedirect.com/science/article/pii/S0301479785700692
Hoeting, J.A.; Madigan, D.; Raftery, A.E.; Volinsky, C.T., (1999). Bayesian model averaging:
A Tutorial. Stat. Sci., 14(4): 382-401 (20 pages).
http://www.jstor.org/stable/2676803

118
 Johnson, J.W.; LeBreton, J.M., (2004). History and use of relative importance indices in
organizational research. Organ. Res. Method., 7(3): 238-257 (20 pages).
http://journals.sagepub.com/doi/abs/10.1177/1094428104266510
Karpušenkaitė, A.; Denafas, G.; Ruzgas, T., (2016). Forecasting hazardous waste generation
using short data sets: Case study of Lithuania. Science–future of Lithuania/Mokslas–lietuvos
ateitis, 8(4): 357–364 (8 pages).
http://www.mla.vgtu.lt/index.php/mla/article/view/951
Kolekar, K.A.; Hazra, T.; Chakrabarty, S.N., (2016). A review on prediction of municipal
solid waste generation models. Procedia Environ. Sci., 35: 238-244 (7 pages).
http://www.sciencedirect.com/science/article/pii/S1878029616301761
Lebersorger, S.; Beigl, P., (2011). Municipal solid waste generation in municipalities:
quantifying impacts of household structure, commercial waste and domestic fuel. Waste Manage.,
31(9-10): 1907-1915 (9 pages).
http://www.sciencedirect.com/science/article/pii/S0956053X11002583
Lebersorger, S.; Schneider, F.; Hauer, W., (2003). Waste generation in households–models in
theory and practical experience from a case study of multifamily dwellings in Vienna.
Proceedings in Sardinia, (4 pages).
Madigan, D.; Raftery, A.E., (1994). Model selection and accounting for model uncertainty in
graphical models using Occam's window. J. Am. Stat. Assoc., 89(428): 1535-1546 (12 pages).
http://www.tandfonline.com/doi/abs/10.1080/01621459.1994.10476894
Mbande, C., (2003). Appropriate approach in measuring waste generation, composition and
density in developing areas. J. S. Afr. Inst. Civil Eng., 45(3): 2-10 (9 pages).
https://journals.co.za/content/civileng/45/3/EJC26944
Memarianfard, M.; Hatami, A.M.; Memarianfard, M., (2017). Artificial neural network
forecast application for fine particulate matter concentration using meteorological data. Global J.
Environ. Sci. Manage., 3(3): 333-340 (8 pages).
http://www.gjesm.net/article_23079_e3b575506205de32a43eea8e244ad182.pdf
MOC, (2010). QCVN 07:2010/BXD National Technical Regulation Chapter 9: Solid waste
collection, separation, transportation, treatment system and public toilet, Hanoi, Vietnam.
http://cucqlxd.gov.vn/media/documents/qcvn_07_2010_cong_trinh_ha_tang_ky_thuat.pdf
MOC, 2016. QCVN 07:2016/BXD National Technical Regulation Chapter 9: Technical
Infrastructure Works, Solid Waste Treatment and Public Toilet.
http://britec.com.vn/source/QCVN/QCVN%2007-1-2016.pdf
Monavari, S.; Omrani, G.; Karbassi, A.; Raof, F., (2012). The effects of socioeconomic
parameters on household solid-waste generation and composition in developing countries (a case
study: Ahvaz, Iran). Environ. Monit. Asses., 184(4): 1841–1846 (6 pages).

119
 https://link.springer.com/article/10.1007%2Fs10661-011-2082-y?LI=true
Ngoc, U.N.; Schnitzer, H., (2009). Sustainable solutions for solid waste management in
Southeast Asian countries. Waste Manage, 29(6): 1982-1995 (14 pages).
http://www.sciencedirect.com/science/article/pii/S0956053X0800442X
Nguyen, D.L.; Hoang, M.G.; Bui, X.T., (2013). Challenges for municipal solid waste
management practices in Vietnam. Waste Technol., 1(1): 17-21 (5 pages).
https://www.researchgate.net/publication/287704977_Challenges_for_municipal_solid_waste
_management_practices_in_Vietnam
Noori, R.; Karbassi, A.; Salman Sabahi, M., (2010). Evaluation of PCA and Gamma test
techniques on ANN operation for weekly solid waste prediction. J. Environ. Manage., 91(3):
767-771 (5 pages).
http://www.sciencedirect.com/science/article/pii/S0301479709003624
Oumarou, M.B.; Dauda, M.T.; Abdulrahim, A.; Abubakar, A.B., (2012). Municipal solid
waste generation, recovery and recycling: A case study. World J. Eng. Pure Appl. Sci., 2(5): 143-
147 (11 pages).
http://rrpjournals.org/wjepas/en_wjepas_vol_2_iss_5_pg_143_147.pdf
Parizeau, K.; Maclaren, V.; Chanthy, L., (2006). Waste characterization as an element of
waste management planning: Lessons learned from a study in Siem Reap, Cambodia. Resour.
Conserv. Recyc., 49(2): 110-128 (19 pages).
http://www.sciencedirect.com/science/article/pii/S0921344906000437
Prades, M.; Gallardo, A.; Ibàñez, M.V., (2014). Factors determining waste generation in
Spanish towns and cities. Environ. Monit. Assess., 187(1): 4098-4105 (8 pages).
https://link.springer.com/article/10.1007/s10661-014-4098-6
Qu, X.-y.; Li, Z.-s.; Xie, X.-y.; Sui, Y.-m.; Yang, L.; Chen, Y., (2009). Survey of composition
and generation rate of household wastes in Beijing, China. Waste Manage., 29(10): 2618-2624 (7
pages).
http://www.sciencedirect.com/science/article/pii/S0956053X09002074
Raftery, A.E.; Madigan, D.; Hoeting, J.A., (1997). Bayesian model averaging for linear
regression models. J. Am. Stat. Assoc., 92(437): 179-191 (3 pages).
http://www.tandfonline.com/doi/abs/10.1080/01621459.1997.10473615
Shamshiry, E., Bin Mokhtar, M., Abdulai, A., (2014). Comparison of artificial neural network
(ANN) and multiple regression analysis for predicting the amount of solid waste generation in a
tourist and tropical area—Langkawi Island. Pproceeding of International Conference on
Biological, Civil, Environmental Engineering (BCEE): 161-166 (6 pages).
http://iicbe.org/upload/2201C0314099.pdf

120
 Sukholthaman, P.; Chanvarasuth, P.; Sharp, A., (2015). Analysis of waste generation variables
and people’s attitudes towards waste management system: a case of Bangkok, Thailand. J. Mater.
Cycle. Manage.: 645–656 (12 pages).
https://link.springer.com/article/10.1007/s10163-015-0456-x
Thøgersen, J., (1996). Wasteful food consumption: Trends in food and packaging waste.
Scand. J. Mgmt., 12(3): 291-304 (15 pages).
http://www.sciencedirect.com/science/article/pii/0956522196000115
Thanh, N.P.; Matsui, Y., (2011). Municipal solid waste management in Vietnam: Status and
the strategic actions. Int. J. Environ. Res., 5(2): 285-296 (12 pages).
https://journal.ut.ac.ir/article_314_736df893be683ae70e91805e3c1fb29e.pdf
Thanh, N.P.; Matsui, Y.; Fujiwara, T., (2010). Household solid waste generation and
characteristic in a Mekong Delta city, Vietnam. J. Environ. Manage., 91(11): 2307-2321 (5
pages).
http://www.sciencedirect.com/science/article/pii/S030147971000174X
Thompson, B., (1995). Stepwise Regression and Stepwise Discriminant Analysis Need Not
Apply here: A Guidelines Editorial. Educ. Psychol. Meas., 55(4):525-534 (10 pages)
http://journals.sagepub.com/doi/abs/10.1177/0013164495055004001#articleCitationDownload
Container
van de Klundert, A., Anschütz, J., Scheinberg, A., (2001). Integrated sustainable waste
management: the concept. Tools for decision-makers. experiences from the urban waste
expertise programme (1995-2001). WASTE.
http://www.waste.nl/sites/waste.nl/files/product/files/tools_iswm_concept_eng1.pdf
Vesely, S.; Klöckner, C.A.; Dohnal, M., (2016). Predicting recycling behaviour: Comparison
of a linear regression model and a fuzzy logic model. Waste Manage., 49: 530-536 (7 pages).
http://www.sciencedirect.com/science/article/pii/S0956053X15302580
Wang, M.C.; Bushman, B.J., (1998). Using the normal quantile plot to explore meta-analytic
data sets. Psychol. Method., 3(1): 46-54 (9 pages).
https://pdfs.semanticscholar.org/fd06/b32eb15d004a1ce32075c4cbc9723b7797b3.pdf
Wilk, M.B.; Gnanadesikan, R., (1968). Probability plotting methods for the analysis for the
analysis of data. Biometrika, 55(1): 1-17 (17 pages).
https://academic.oup.com/biomet/article-abstract/55/1/1/262814
Zurbrugg, C.; Gfrerer, M.; Ashadi, H.; Brenner, W.; Kuper, D., (2012). Determinants of
sustainability in solid waste management--the Gianyar Waste Recovery Project in Indonesia.
Waste Manage., 32(11): 2126-2133 (8 pages).
http://www.sciencedirect.com/science/article/pii/S0956053X12000256

121
6. MODM Model

6.1 Introduction

Waste management is one of the most important issues concerning environmental protection
and resources conservation (Costi et al., 2004). SWM in developing countries has received
increasing attention from researchers and policy makers hoping to follow a sustainable approach
to waste management and to integrate strategies that will produce the best practical option.

That is a difficult task, as it is essential to take into account economic, technical, and
environmental aspects properly as well as the social agreement and every society has several
goals for implementing SWM plans depending on some circumstances and issues. For example,
the Hoi An city facing a landfill space problem may set a target to reduce the quantity of waste
sent to landfill disposal or may consider source decrease, waste diversion through recycling and
volume reduction alternatives such as converting waste to energy. The most sustainable choice,
however, is often not clear. Important questions need to be answered is how MSW flows should
be managed and what are appropriate treatment technologies for a sustainable waste management
plan.

Optimization helps to find the answer that yields the most reliable solution of the problem
such as the lowest cost or highest profit, lowest emission/waste or highest product; or achieves
the lowest discomfort or highest happiness. In recent years, many works have presented useful
and comprehensive decision-making models which are both significantly close to reality and
computationally tractable in order to help planners manage solid waste treatment and disposal in
urban area. However, some model does not take into account normative, environmental and
technical aspects (Y. H. Chang et al., 1998; Fiorucci et al., 2003).

In Vietnam, the municipal authority which has responsibility for SWM focuses primarily on
organizational aspect such as improvement of municipal solid waste management quality or
upgrading of waste management tools. However, in reality, waste management also consists of
economic sector, comprising a variety of interlinked actors, activities and commodities (Lyeme
et al., 2017). Thus, there are always conflicts exist among various stakeholders involving in
waste management system. A holistic approach for modelling waste management in Viet Nam

122
for DSS to help decision makers is essential. Nevertheless, the application of MODM in Viet
Nam has been left so far compared to other countries; there have been no studies in waste
management modelling conducted in Viet Nam.

In this framework, the purpose of this chapter is that of considering a quite general model,
which is comprehensive of all technical, economic, normative, and environmental aspects
concerning the management of MSW. Specifically, the proposed system can help decision
makers choose the solution to manage the MSW stream and the capacity of appropriate treatment
plants, by a careful analysis of the waste generation and composition, physical and chemical
characteristics. The problem to be solved is in a category of nonlinear mathematical
programming problem. In this way, the optimal configuration of the system (from the physical
and technological point of view) can be determined, as well as the optimal flows among various
plants in the system. Environmental impacts are carefully represented, social acceptance is
optimized due to the participation of various DMs (including citizens) in choosing the system
and intensive discussion in decision-making process. The interactive approach for solving the
model was applied to achieve social acceptance optimization is Reference Point Method.

6.2 Assessment to the desires of society

A questionnaire survey of 18 local experts with a face-to-face interview was conducted in

2015. The purpose of the survey is to assess the requirements of different stakeholders who
involved in waste management in Hoi An. The local experts who were interviewed including
authorities, waste managers, stakeholders, citizens’ representatives, etc. which will be explained
in Table 6.1. The main focus of the survey is to find out the most important objectives for the
waste management system in the city, the potential treatment technologies for the current local
conditions, and how the waste stream should be managed, etc. The information provided based
on the knowledge and self-evaluation of local experts who may not have comprehensive
knowledge about the integrated waste management but have solid experience and profound
understanding of the local system. The final results of this survey will contribute to optimization
model of Hoi An waste management.

Table 6.1 Experts and stakeholder group being interviewed

123
 Expert Group Explanation of the expert’s role
A1 Authority People working in City Committee and Department
A2 of Environment and Natural Resource, in charge of
waste management and environmental management
in Hoi An
S1 Contractor Waste recycler works privately
S2 People working HAPSC the company in charge of
S3 solid waste collection and treatment in Hoi An city
S4
C1 Citizen The expert C1 to C11 are the people who live in Hoi
C2 An city, working as a monitor for waste management
C3 activities in their communities. They had a degree in
C4 environmental management. Each expert represents
C5 for one community in Hoi An including Cam Thanh,
C6 Cam Ha, Cam An, Minh An, Tan An, Cam Kim,
C7 Cam Pho, Cam Nam, Cua Dai, Son Phong, Thanh
C8 Ha.
C9
C10
C11
C12 Expert C12 is a researcher living in Hoi An city and
conducting research in waste management locally
and internationally.
6.2.1 Methodologies

The “degree of consensus” denotes the degree of similarity of preference among

stakeholders and the “consensus results” signifies the average preference of all stakeholders. To
calculate the consensus results and the degree of consensus, consensus analysis model (CAM) is
applied (Hung et al., 2007). Fuzzy set theory is applied to deal with linguistic variables in CAM,
which are employed to identify stakeholder attitudes and opinions. The CAM procedures are
presented as following:

Determine linguistic variable for alternative preference

Five levels (very good, good, moderate, bad, and very bad) and (very important, important,
moderate, not important, not necessary) are used as linguistic variables to assess the
stakeholders’ opinion (Zimmermann, 2011).

Fuzzy subjective decision matrix

The Fuzzy subjective decision matrix for each stakeholder can be established as follows:
124
 ~
x ik  [ xikl , xikm , xikr ]

xikl  min{xijk
l
}j, k

xikm  geomean{xijk
m
}j, k

xikr  max{xijk
r
}j, k (6.1)

Where xikl ( xikr ) denotes the left (right) value (the minimum (maximum) value in this study)

of the fuzzy number of preference for alternative k judged by stakeholder i, xikm presents the
medium value (the geometric mean value) of the fuzzy number of preference for alternative k
l r
judged by stakeholder i; xijk ( xijk ) denotes the left (right) value (min/max value) of the fuzzy

number of preference for alternative k judged by expert j in the stakeholder group i.

The fuzzy decision matrix

The fuzzy decision matrix can be established be de-fuzzifying the data using the centroid
method (Solymosi et al., 1986). The fuzzy decision matrix is presented in the equation 6.2, where
fik is the defuzzified value of stakeholder i for alternative k.

 f11 ... ... ... f 1n 

     
 
  f ik   (6.2)
 
     
 f m1 ... ... ... f mn 

Calculating the consensus results for the alternatives

The consensus results for each option can be expressed as the mean defuzzified value of the
triangle fuzzy number for stakeholders as follows:

CR = [ f 1 , f 2 ,... f k ,..., f n ] (6.3)

Where f k is the average of the defuzzified value of all expert in all stakeholders for
alternative k.

125
 Obtaining consensus degree for alternatives

After the consensus results for the alternatives are determined, the consensus degree for
these alternatives is also defined. The calculation of the degree of consensus is conducted in
three steps:

Firstly, semantic standard deviation distance for each option (sdk) is calculated as the
following equation:

( f ik  f k )2
sd k  i
(6.4)
m 1

Then, the semantic distance is normalized. The normalized semantic distance is calculated
by the following equation:

sd k
d knorm  (6.5)
sd *

Where sd* is the maximum of all possible sd.

Lastly, the degree of consensus is obtained for each alternative (CDk). The consensus degree
for each alternative is calculated using the following equation:

CDk = 1- d knorm (6.6)

According to Hung et al. (2007), the degree of compromise for consensus decreases with the
larger numbers of stakeholders. Therefore, as the number of possible situations is known, if the
degree of consensus exceeds a certain probability, then it can be said that a compromise (or
acceptable level) has been reached. Table 6.2 presents the degree of compromise for consensus
with different numbers of stakeholders.

Table 6.2 Degree of compromise for consensus with various number of stakeholders

Number of stakeholders 3 4 5
Degree of compromise for consensus 0.500 0.426 0.385

126
6.2.2 Results of the expert's survey

6.2.2.1 The objectives of a sustainable solid waste management

Table 6.3 presents the results of the average defuzzified value of all stakeholders for the
objectives of the sustainable waste management system.

Table 6.3 The CR value of general targets for sustainable waste management

Stakeholders Objectives Economic Objectives Environmental Objectives

Economic Environmental Social Cost Benefit Landfill GHG Emission
Authority 0.750 1.000 0.625 1.000 0.500 1.000 0.375 0.375
Contractor 0.875 0.75 0.625 1.000 0.000 0.938 0.188 0.542
Citizens 0.542 0.896 0.813 0.667 0.167 0.854 0.25 0.618
CR value 0.722 0.882 0.688 0.889 0.222 0.931 0.271 0.512

Experts agreed that three objectives including economic affordability, environmental

efficiency and Social acceptance are essential for the sustainable management (CR>0.500). In
Economic objective, minimizing cost should be focused in the model rather than benefit as the
evaluation of all interviewees. Local experts understand the situation of real-world that it is hard
to earn benefit from waste management in a developing city (Hoi An). GHG emission reduction
got less attention according to the survey (CR=0.271) in environmental objectives than landfill
reduction and emission reduction. This reflects correctly the real situation in Hoi An city when
the land resource for landfill is limited, the current landfill has been full and caused pollution.
Moreover, the global warming reduction from waste management got the lack of understanding
and attention in developing countries like Viet Nam.

Figure 6.1 Present the degree of consensus for chosen objectives, it shows that there are a
high level degree of consensus for chosen objectives. The consensus level of objective among
stakeholders as well as in each stakeholder group for each objective. Thus, we can consider Cost,
Landfill and Emission as objectives in sustainable waste management model.

6.2.2.2 Treatment Alternatives

The same method is applied for evaluating the stakeholder consensus of treatment options.
Stakeholders will answer which treatment technology is the most suitable for the local

127
socioeconomic conditions, technical level conditions based on their own experience and
knowledge.

1.000
Authority Minimize Cost 0.808
1.000
0.508
1.000
Contractor 0.857 Minimize Landfill 0.927
0.751

0.766 0.823

Citizen 0.708 Minimize Emission 0.876

Red: DC of Authority; Blue: DC of Contractor; Black: DC of Citizen; Green: Overall DC for the objective

Figure 6.1 The degree of compromise for consensus (DC value)

Table 6.4 Consensus results and Degree of consensus for treatment options

INCE INC COM ADE AD HT LF REC

Authority 1.000 0.625 0.750 0.750 0.375 0.750 0.500 0.875
Contractor 0.813 0.875 0.625 0.313 0.188 0.563 0.688 0.938
Citizens 0.813 0.438 0.688 0.729 0.375 0.729 0.583 0.896
CR value 0.875 0.646 0.688 0.597 0.313 0.681 0.590 0.903
DC value 0.892 0.781 0.938 0.753 0.892 0.897 0.906 0.968

6.3 Modeling waste management system

6.3.1 Waste flow analysis

The aim of the dissertation is that of considering a general model, which is considering all
technical, economic, and environmental aspects of MSW management. The proposed model can
help decision makers choose the solution to manage the waste stream, the capacity of potential
treatment facilities, by a careful analysis of waste material flow. Nonlinear mathematical
programming will be applied to solve the model, and the optimal configuration of the system can
be determined, as well as the optimal flows among the various plants in the system. A non-linear
model was developed in General Algebraic Modeling System (GAMS) software. The model has

128
277 variables and 191 constraints with 1255 Jacobian elements, 3 of which are nonlinear, and 3
nonlinear variable. The size of the program is 4 MB and it takes 0.5 seconds to solve by GAMS.
The waste composition analysis results (in section 4) indicated that there is no significant
difference in waste composition among areas in HAC, author, therefore waste from different
areas was not considered.

This model based on the waste flow analysis approach. W denotes the daily waste generated
in HAC, Ri,k represents the amount of waste fraction i which will be transported to treatment
facility k. It should be noted that recycling is possible for 6 kinds of materials including paper,
cardboard, plastic, plastic bags, metal and glass, those can be separately collected by informal
recycling such as scavengers collects from treatment plants, from collection points or buying
directly from households. Also, the organic material (Food and Garden waste) is recycled at
home by feeding animals or home composting by few households. According to Chu (2014) total
amount of waste recycled was about 5% and 5-10% of total MSW for organic materials and
recyclable waste, respectively. The remaining fraction of MSW (after separated for recycling at
home) is separated into 2 types of waste degradable and nondegradable, then collected and
transported to treatment facilities: Composting (COM), Anaerobic Digestion (AD), Incineration
(INC) and Incineration with electricity production (INCE). Those were chosen as potential
treatment technologies for HAC by eighteen interviewed DMs.

Each treatment facility generates their own residuals which need to be treated. The waste
flow WRADi,k’ is the residual of material i after being processed by anaerobic digestion process
(including primary treatment and anaerobic process), which go to COM, or INC, or INCE or LF
for further treatment. Similarly, WRCOM i,k’ is the residual of fraction i after being processed in
Composting facilities, that will be treated at incineration or landfill later. WRINC is the total
amount of residual from incinerator being landfilled. Products produced from composting plant
and energy generated from Incineration will be sold to the market.

To obtain the data of material flow after each treatment facility, LCA and LCI study should
be conducted for that treatment plant. However, it also should be noted that the direct
implementation of LCA for each treatment facility is complicated, costly and time-consuming
(Chiu et al., 2008). However, the exact LCI calculation or environmental impact is not the main

129
purpose of this study. Thus, we applied data from various reliable scientific reports, articles. Data
and sources are presented in table 6.5 and 6.6.

Figure 6.2 The waste flow structure of the model

6.3.2 Waste flow modelling

The input and output flow of different treatment facilities are described in the model. The
indices that are used in the mathematical model are introduced as follows. Waste types i is
generated from residents in ward j and transported to treatment facility k. Moreover, the index p
denotes the type of pollutants emitted from treatment technology k. The list and explanation of
sets, parameters and variables used in the model are shown in Table 6.5. The Eq. 6.7-6.8
emphasise that the available waste i should be allocated to different treatment facility k. The set
of Eq 6.9-6.25 describes the waste flow going in and going out for different treatment facility k.

130
 Table 6.5 Sets, parameters and variables

Set
i Waste type: Food (Fd), Garden (Gd), Plastic (Pl), Plastic bags (Pb), Cardboard
(Cb), Paper (Pp), Metal (Mt), Glass (Gl), Combustible waste (Co), Inert (In),
hazardous waste (Hz).
id A subset of i denotes the degradable waste fraction including Fd and Gd
in A subset of i denotes the nondegradable waste materials
k Solid waste treatment options for the city including Composting (COM),
informal recycling (IRE), home treatment (HT), anaerobic digestion with
energy recovery (ADE), Incineration (INC), Incineration to energy (INCE),
Landfill (LF)
k’ A subset of k, present for treatment option which can treat the waste residual
from other waste disposal facilities including COM, INC, INCE, IRE and LF
p Pollutants emitted set including CH4,CO2, SO2, NOx, Heavy Metal, VOC,
N2O.
Parameters
αi Percentage of waste i in MSW
βi,k The amount of residual generated from treatment technology k by treating 1
ton of waste i (Table 6.6).
m
λ i,k The amount of product produced from treatment technology k by treating 1
ton of waste I (Table 6.6).
Amount of residual generated from incineration of 1 ton of anaerobic
sres, INC digestion sludge (equal to 0.4) (FHWA, 2016)
sres,COM Amount of residual generated from incineration of 1 ton of anaerobic
digestion sludge (assumption equal to 0)
Amount of compost product generated from composting of 1 ton of anaerobic
spro,COM
digestion sludge
LHVsludge Lower heating value of sludge generated by AD (9.7Mj/kg)(Khan et al., 1991)
LHVi Lower heating value of waste i
ek Electric consumption of treatment k for a ton of input waste
k The efficiency concerning of energy production concerning heat produced by
combustion of waste in incinerator and biogas from AD.
μp,k Amount of pollutant p emitted from treatment process k by treating 1 ton of
input waste
ckcap , ckop Investment and operation cost of 1 ton of waste to treatment facility k.
c trans Transportation cost of 1 ton waste to treatment facility k
bcom Price for 1 ton of compost fertilizer
benergy Price for 1 kWh of electric by Waste to Energy
Variables
Ri ,k Amount of waste i transported to treatment facility k
RADi,IRE Amount of recyclable waste type i collected from the input waste of AD
facility by scavengers

131
RCOMi,IRE Amount of recyclable waste type i collected from the input waste of
Composting facility by scavengers
RINCi,IRE Amount of recyclable waste type i collected from the input waste of
Incineration by scavengers
WRADi ,k ' Amount of residual sludge generated after treatment from AD facility type i
to facility k’
WRCOM i ,k ' Amount of residual generated after treatment from Composting facility type i
to facility k’
Ck Amount of waste will be processed in facility k

Waste flows

W   i   Ri ,k i, k (6.7)
k

The amount of recyclable waste after

C IRE   Ri , IRE   RADi , IRE   RCOM i , IRE   RINC ie, IRE  WRCOM i , IRE  WRADi , IRE
i i i e i i i

i recycle (6.8)

Constraints for recyclable waste recycled informally

Ri , IRE  RADi , IRE  RCOM i , IRE  RINCie, IRE  0

with i = Fd, Gd, Co, In, Hz (fraction i is not recyclable waste) (6.9)

The Eq 6.10 is a constraint indicating that the total amount of recycled waste is less than
10% of total MSW.

R
i
i , IRE   RADi , IRE   RCOM i , IRE   RINC ie, IRE  WRCOM i , IRE  WRADi , IRE  0.1  W ,
i i e i i i

→ CIFR  0.1  W

i= recyclable waste (6.10)

Home treatment for degradable waste

CHT   Ri , HT i  Fd , Gd (6.11)
i

The amount of degradable waste treated at home less than 5% of total MSW

132
 C HT  0.05  W (6.12)

R
i
i , HT  0 , j , i  Fd , Gd (6.13)

Anaerobic Digestion

C AD   Ri , AD i (6.14)
i

WRADi , k '  ( Ri , AD  RADi , IFR )  i , AD i = Fd, Gd, Pp, Cb,

k’ =COM, INC, INCE, LF (6.15)

Sludge generated from anaerobic digestion from Food and Garden waste will be transported
to other treatment such as composting, incineration, landfill.

Biogas product

BO   ( Ri , AD  RADi , IRE )  im, AD (6.16)

i

Composting

CCOM   Ri ,COM   WRADi ,COM (6.17)

i i

WRCOM
k'
i ,k '  ( Ri ,COM  RCOM i , IFR )   i ,COM  WRADi ,COM  sres ,COM , i (6.18)

Waste generated from composting plant not re-treated at composting facility

WRCOM
i
i ,COM 0 (6.19)

Compost product:

CF   Ri ,COM  im,COM   WRADi ,COM  spro ,COM (6.20)

i i

Incineration

Ck   Ri , k   WRADi , k   WRCOM i , k i ; k =INC, INCE (6.21)

i i i

133
 WRINC    ( Ri ,k  RINC ie, IFR )   i ,k    WRCOM i ,k '   i ,k '    WRAD i ,k ' sres , INC (6.22)
k i k' i k' i

Landfill

C LF   Ri ,LF  WRADi ,LF  WRCOM i ,LF  WRINC (6.23)

i i i

Table 6.6 Product and residual rate of treatment alternatives

Composting (ton per ton) AD (ton per ton) Incineration

Waste fractions Compost λ i,k
m
Residual βi,k Biogas λ i,k
m
Residual βi,k LHV (MJ/t)(d) Residual(d)

Food (Fd) 0.25 (a) 0.059 (a) 0.130(c) 0.2(c) 1.792 0.194
Garden (Gd) 0.649 (b) 0.0793 (b) 0.039(c) 0.2(c) 5.625 0.191
Plastic (Pl) 0 (e) 1(e) 0(e) 1(e) 29.170 0.047
Plastic bag (Pb) 0 (e)
1 (e)
0 (e)
1 (e)
24.888 0.047
Cardboard 0(e) 1(e) 0.144(c) 0.2(c) 10.932 0.125
Paper 0 (e)
1 (e)
0.129 (c)
0.2 (c)
11.707 0.114
Metal 0(e) 1(e) 0(e) 1(e) 0(e) 1(e)
Glass 0(e) 1(e) 0(e) 1(e) 0(e) 1(e)
Combustible 0(e) 1(e) 0(e) 1(e) 19.818 0.048
Inert 0 (e)
1 (e)
0 (e)
1 (e)
0 (e)
1(e)
Hazardous 0(e) 1(e) 0(e) 1(e) 23.505 0.038
Source: (a)
(Kim et al., 2010) , (b)
(Jacob K Andersen et al., 2010),
(c)
Database of waste management technologies (EPEM),(d) Sample analysis by author, (e) assumption

Energy consumption and production

Energy consumption for treatment facilities

EC k   Ri ,k  ek , k (6.24)
i

The amount of energy consumption per ton of waste (ek) is various depend on numerous
factors such as the facility technology, capacity, ages or managerial capacity, etc. Thus, the
author used available data from existing LCA study as references for energy consumption rate of

134
different treatment techniques. The power consumption rate parameters are presented in Table
6.7.

Table 6.7 Energy consumption coefficients

COM AD INC INCE LF HT IFR

ek (Kwh per ton) 62(1) 57(2) 70(3) 70(3) 7(4) 0 0
(1) (2) (3) (4)
Sources: (Lee et al., 2007), (Khoo et al., 2010), (Tan et al., 2006), (Ngoc, 2016)

Energy produced by AD

EPAD   AD  BO  LHVbiogas / f (6.25)

Energy produced by INCE

EPINCE   INCE  (  ( Ri , INCE  RINCi , IFR  RINCEi , IFR )  LHVi  WRCOM i , INCE  LHVi 
i i
(6.26)
 SAD
i
i , INCE  LHV sludge
  (WRADi , INCE  SADi , INCE )  LHVi ) / f
i

EPk  0 , k  AD, INCE (6.27)

Where,  AD and  INCE is the efficiency concerning energy production concerning heat

produced by combustion. An assumption of 40% and 27% are made for  AD and  INCE ,
respectively.

f : energy conversion factor from MJ to Kwh, equal to 3.6 MJ/Kwh

Emission from treatment facilities:

EM k   C k   p ,k (6.28)
p

Table 6.8 Emission factors  p,k

 p,k COM AD INC HT(2) LF

CH4 4(3) 1(3) 0(3) 2 500(3)
CO2 291 227 1000 252 300
NOx - 0.188 1.6 - 0.68
SO2 - 0.03 0.042 - 0.053

135
VOC 24 2.1 0.008 24 6.4
Heavy metal(1) 0.00265 0.0027 0.01253 0 0.00007
N2O 0.3(3) 0(2) 0.05(2) 0.45 0
(1) (2)
Source: (DEFRA, 2004), (J. K. Andersen et al., 2011; EPEM), (3) (IPCC, 2006) and
(Harijani et al., 2017)
Total cost for waste management

According to the general plan of the city, there is only one area for a solid waste treatment
complex located in Cam Ha Ward. Therefore, the waste collection and transportation fee in this
study is calculated based on the financial data provided by HAPWS Ltd.Co, via a personal
interview with the vice-director. Hoi An is a small city, and the distance from the city centre to
the waste treatment complex is about 10km. Thus, in this study author will not consider the
distance from different places to waste disposal plant and the waste collection and transportation
cost for 1 ton of MSW (ctrans) is about 275.000VND (equivalent to 13UD per ton) (Ngoc, 2016).

TC   Ri ,k  c trans , i ,k ≠ HT, IFR (6.29)

k i

The unit costs of investment (capital) and operation (variable) for treatment facilities in Viet
Nam are regulated in Decree 32/QD-BXD (MOC, 2012). However, due to the lack of
information in AD with energy recovery in Viet Nam, author assume the cost for investment and
operation of AD and INCE technology, also assume the lifetime of the system is 20 years.
Capital and variable cost will be presented in Table 6.9, the rate used is for the small scale
treatment facility. Thus, the capital cost per day for each facility is presented as follows:

CC k   Ri ,k  c cap
k /(30  365) (6.30)
i

COPk   Ri ,k  c op
k (6.31)
i

Table 6.9 The capital cost and variable cost of treatment facilities

Cost (US$ per ton.day) COM AD INC INCE LF HT IFR

Investment 21140 34225 27307 38000 9899 0 0
Operation 12 15 18 20 6 0 0
Source: (MOC, 2012)

Revenue from selling composting product

136
 Composting plant: BCOM  CF  bcom (6.32)

Revenue from selling electricity product

Benergy  (  EPk   EC k )  benergy moreover, Benergy  0 when

k k
 EP   EC
k
k
k
k 0 (6.33)

Vietnamese Government supports energy production from renewable energy power plant.
The buyer is responsible for purchasing the entire power generated from projects treating solid
waste with the price of 2,114 VND/kWh (equivalent to 10.05 US cents/kWh) for power
generating directly from burning solid waste (Decision:31/2014/QĐ-TTg, 2014). The price for 1
ton of composting fertilizer in HAC is 7 US$ per ton as the result of author’s survey in 2015.

Positive constraints

Ri,k, RAD, RCOM, RINC, RINCE, WRADi,k, WRCOMi,k, Ck ≥0 (6.34)

6.4 Single objective optimization model

6.4.1 Objective function

The waste management system needs to operate at a cost accepted by the community
including local government, stakeholders and citizen. Thus, minimising cost or maximising
benefit of the whole system is widely used in previous studies of optimisation model of waste
management to consider the economic aspects (N.-B. Chang et al., 1996; Galante et al., 2010;
Harijani et al., 2017).

In this study, authors considered estimating the total cost of the system including
transportation cost, capital cost, variable cost and the revenue from selling recycle material and
energy products. The goal of gaining economic profit from waste management is difficult to
achieve, especially in a developing country like Vietnam. Also, as the result of the expert’s
survey, Therefore, the optimisation model in this research is developed to minimise the cost of
the municipal waste management system.

The objective function developed in this model is presented as the following equation

Minimize Z   TC
k
k  CCk  COPk  Bcom  Benergy (6.35)

137
6.4.2 Scenarios development

6.4.2.1 Landfill reduction initiatives

One of the most significant problem in HAC currently is the landfill problem. A major
amount of waste has been dumped in an open landfill for a long time, and it has become overload
and full from 2015. Thus, most of the waste generated in the city now has being transported and
dumped in Nui Thanh landfill, which is about 70km far away from the city centre. In addition,
due to the rapid growth of tourism industry, a land resource for landfill is limited. Therefore,
landfill reduction target has the most concern from authorities and waste management experts as
the results of expert’s survey. However, reducing waste-to-landfill needs a wise initiative with a
clear plan of action. Thus, this study will evaluate different targets of landfill reduction to have a
transparent view of system capacity in local conditions to support decision makers in creating a
plan for landfill reduction. Three targets of landfill reduction including 50%, 25% and 10% of
MSW being landfilled will be set for calculation of minimizing system cost, GHG emission and
various pollutants emission.

A constraint of landfill target will be added to the model, which is described in the equation
6.36

C LF  LT (6.36)

LTR is landfill target reduction, equals to 0.5, 0.25 and 0.1

6.4.2.2 Sensitive analysis for waste separation efficiency

The efficiency of waste sorting at source affects significantly to system effectiveness

including technical and economic aspects. Hoi An is the first, and the only city in Vietnam has
succeeded in separation program, but the rate of separation is low and varied (approximately 70-
85% as reported in 2014-2015). Sensitive analysis for waste separation efficiency provides
information about the effect of sorting waste to the system cost and pollution emission, which
might support decision makers in controlling and evaluating a reasonable separation rate for
optimizing the system. The method approach in this study for sensitive analysis of waste
separation efficiency is scenarios analysis. Thus, the author will estimate the result of a model
for different sorting efficiency as assumed: 70%, 80%, 90% and 100% for completely separated.

138
 Two constraints for the efficiency of waste separation at source for nondegradable waste and
degradable waste are formulated in the model:

Non-degradable waste not separated and transported to the biotreatment plant (AD and
COM)

Rin , AD  Rin ,COM  (1  SE )  W   in (6.37)

Degradable waste not separated and transported to the INC, INCE and LF

Rid ,INC  Rid ,INCE  Rid ,LF  (1  SE )  W   id (6.38)

With three values for LT and four waste separation rates, 12 scenarios are developed as
presented in Table 6.10

Table 6.10 Scenario description

SC1 SC2 SC3 SC4 SC5 SC6 SC7 SC8 SC9 SC10 SC11 SC12
Landfill target 50 50 50 50 25 25 25 25 10 10 10 10
(LT), %
Separation rate 100 90 80 70 100 90 80 70 100 90 80 70
(SE), %
Thus, the model includes Variable, parameters and constraints was explained above.

6.4.3 Results of Scenarios analysis by single optimization model

6.4.3.1 Economic assessment

Figure 6.3 Total Cost and Emission fo 12 Scenarios

139
 The Single optimization model searches for the optimal solution of minimizing cost for each
examined Scenarios. Figure 6.3 presents the results of cost and total emission from each scenario.
The results show that Cost is proportional with the landfill target, the higher landfill reduction
target, the higher cost for the system. Also, the comparison among scenario of a group with the
same landfill target (SC1-SC4 (LT:50%), SC5-SC8 (LT:25%) and SC9-SC12(LT:10%)) shows
that the separation efficiency affects significantly to the system cost. The lower separation
efficiency leads to higher cost of the system. The blank result of SC12 means that the model
show infeasible solution of the model with the constraint of 10% waste-to-landfill and low
separation efficiency of 70%. In another word, The target of 10% waste-to-landfill cannot be
reached if the rate of waste separation at source is lower than70%. For each landfill target, the
lowest cost is found when MSW is separated completely (SC1, SC5 and SC9) and the SC1 has
the lowest cost among all scenarios in the evaluation.

The waste separation has the same effect with emission, a lower rate of waste sorted at
source causes higher emission (comparing among the scenarios with the same LT (SC1-SC4:
50%, SC5-SC8:25% and SC9-SC11:10%). The total emission is proportional with the system
expense which seems not reasonable because it is expected that spending more money can
reduce emission. However, the system emission depends on the technologies used. Figure 6.4
presents treatment alternatives of scenarios as the results of the mode. The landfill is known as a
large producer of CH4 and CO2 and Incineration is the generated highest amount of CO 2 per ton
of waste treated among treatment technologies considered in this study. CH 4 and CO2 have the
largest emission in weight in waste among all pollutants considered in this study. When the
landfill rate is high ( in scenario SC1-SC4), a big amount of MSW waste will be landfilled, less
composted and no incineration. As a result, high amounts of CH4 and CO2 (Figure 6.5) is emitted
from landfill leads to an increase in total emission of the system.

The scenario SC9-SC11 produces a high amount of emission due to the low landfill target of
10%, which causes the system favour incineration more than composting or anaerobic digestion.
The LT of 25% is more reliable in term of emission because the system can favour recycling
technology such as Composting or Anaerobic Digestion, which generated fewer pollutants. The
scenario SC5 has the lowest emission among all evaluated scenarios as a result of the model.

140
Figure 6.4 Treatment options for 12 Scenarios

141
6.4.3.2 Waste management emission

Figure 6.5 Emission of CH4, CO2, NOx, VOC, SO2, N2O and heavy metals

Figure 6.5 shows the emission of each pollutant for each scenario. The CH4 and SO2 are
emitted more when the landfill is favoured (SC1, SC2, SC3, SC4) while CO2, N2O is generated
more when incineration and composting is applied (SC5-SC11).

6.4.3.3 GHG emission and reduction

Figure 6.6 GHG emission of alternative Scenarios

142
 Figure 6.6 compares the GHG emission (GWP applied for a period of 100 years) of 12
alternative scenarios. The results show that the scenario with a high amount of waste to landfill
(SC1, SC2, SC3, SC4) are the largest GHG emission of about 0.622 Mt CO 2-eq per day (SC1) and
0.624 Mt per day (SC4). The slight increase of GHG caused by the increase of waste being
composted and incinerated (which generates CO2) to deal with lower separation efficiency to
ensure the amount of waste-to-landfill being unchanged. However, to reduce the GHG emission,
composting and waste incineration technology should be applied.

6.4.3.4 Waste flow analysis

The result of Scenarios SC5 (LT:25% and SE:100%) was applied for result explanation. The
specified amount of each waste fraction go to each treatment alternative is presented. As a result,
Food waste is favoured for home treatment by citizens which can be used for home composting
or feeding animals (3.5 t/day). 25 ton of Food waste and 7.6 ton of Garden waste are transported
to composting plant whose capacity should be 32.6 t/day. Thus, the plant capacity for
Composting is 32.6 ton per day. The total amount of waste to landfill is 17.532 t/day including
14.86 ton of mixed waste (accounted for 21.2% of MSW waste), 2.1 ton residual of composting
plant and 0.57 ton residual from incineration and 22.44% of MSW is incinerated. The decision
variables and waste stream of MSW system as results of the model for SC5 is showed in Figure
6.7

143
 Figure 6.7 Material Flow of Waste Management System
Landfill target of 25% and separation efficiency of 100%

144
6.5 Multi-objectives optimization decision-making model

6.5.1 Objective functions of the model

According to the CAM analysis for experts consensus. The Economic objective should
minimize the cost of the system rather than benefit gaining. Also, GHG emission reduction got
less attention in environmental objectives than landfill reduction and emission reduction. Thus,
three objective functions modelled in this study are explained as follows:

Minimizing Cost

Authors considered estimating the total cost of the system including transportation cost,
capital cost, variable cost and the revenue from selling recycle material and energy products. The
equation 6.36 of single optimization model was used.

Minimize Z 1   TC k  CC k  COPk  Bcom  Benergy (6.39)

k

Minimizing Emission

According to F.R. McDougall and Hindle (2001), the term “environmentally effective”
means that the waste management system must reduce as much as possible the amount of
pollutant from waste management such as CO 2, CH4, SOx, NOx, heavy metal,… In this study, the
minimisation of the total amount of pollutants is considered as the environmental objective of the
model. Emission of various pollutants in this article including CH 4, CO2, NOx, SO2, N2O, VOC
and heavy metal were calculated according to emission factors reported in previous studies and
reliable scientific reports.

Minimizing Z 2   EM k (6.40)
k

Minimizing waste to landfill

The landfill reduction gain the most attention from local experts (CR value is equal to 0.9)
reflects correctly the real situation in Hoi An city, when the land resource for landfill is limited,
the current landfill has been full and caused pollution. Thus an important objective will be
applied in the study is minimizing the amount of waste to landfill which is presented in the
following equation.

145
 Minimizing Z 3   Ri , LF  WRADi , LF  WRCOM i , LF  WRINC (6.41)
i i i

In the Multi-objective model, we will not compare the landfill reduction target. Thus, the
Equation 6.37 and Equation 6.38 will not be used in the model.

Social Agreement

A sustainable waste management system should be environmentally effective, economically

affordable and socially acceptable (F.R. McDougall and Hindle, 2001). In this study, to assess
the social acceptance, the author will apply The Reference Point Method and its interactive
technique, which was firstly proposed by Wierzbicki (1982) and later successfully implemented
in engineering design, solving environmental management problems as well as other fields of
decision support (Wierzbicki et al., 2000).

6.5.2 Decision-making process approach

The interactive procedure of Decision-making process applied in this study with RPM is
described in Figure 6.8. The MODM method is applied here determines efficient solutions using
interactive procedure with reference point optimization method, which was developed by
Wierzbicki et al. (2000) and implemented in waste management in a study of Minciardi et al.
(2008). The process of decision-making is explained as follows:

Vector x represents the decision variables to be selected within the feasible set Q, and F(x) =
(f1(x), f2(x),.. fm(x)) is a vector function that maps the feasible set Q into the Criterion space Rm.

An outcome vector Z is attainable if it express outcomes of feasible solution, if each element

zj,j∈[1,2,..m] = min j∈[1,2,..m] fj(x) or Z=min(F(x)) for x∈Q.

z j  z min
Value  j  j
plays the role of a normalizing factor of objective value z j . For each
z max
j  z min
j

criterion j, z max
j and z min
j are respectively defined by maximum and minimum values of f j(x)

obtained by performing the corresponding single objective optimization. Thus,  j  [0,1]

(  j  0 when z j = z min
j and  j  1 when z j = z max
j ). The z max
j and z min
j are presented in the trade-off

Table 6.11

146
 Figure 6.8 Decision-making process

The basic procedure of decision making is described as the following. Firstly, the DM
choose a set of aspiration values of objective function, denoted as the initial reference point  j .

Then, a simple form of order-consistent achievement function is created as follows:

m
  min (  j   j )    (  j   j ) with ε is a very small positive parameter (6.43)
1 j  m
j 1

The Equation 6.43 is a prototype of all order-consistent achievement scalarizing function

and is monotone concerning all  j . Thus, all its minima are Pareto-optimal (Wierzbicki et al.,

2000). Lastly, a new solution of decision variables x * is determined by maximizing the

achievement function σ. Figure 6.8 describes the whole procedure of decision making.

An experiment of the MODM process was conducted in 2017. Three colleagues of the
author were invited to play the role of different DMs. The role of each DM as follows: The

147
Decision maker 1 (DM1) represents for Authority, DM2 represents for citizen who generated
waste and the DM3 is from a waste collection and treatment contractors.

Table 6.11 Trade-off table

Single objective Cost (US$/day) Emission (kg/day) Landfill (t/d)

Minimizing Cost (Z1) 1213.4 (min) 50 080 (max) 59.6 (max)
Minimizing Emission (Z2) 2099.8 32 415 (min) 27
Minimizing Landfill (Z3) 3226.2 (max) 38 909 5.36 (min)

6.5.3 Results of MODM

6.5.3.1 Pareto-optimal solutions of decision-making process

After having analyzed the results obtained in the first interaction, DM1 decided quickly with
the second interaction with landfill reduction is the priority objective. DM1 was willing to accept
a worsening in objective Z1 (cost), accepting emission reference point value of 0.3, with the goal
of minimizing waste to landfill (RP=0). The DM2 was more interesting in pollution reduction as
he did not change the RP of Emission (RP=0), he slightly made a trade-off between for Cost and
Landfill to keep minimizing the Emission. The DM3 spent more time to learn the model
behaviour, he made four interactions with the model, trying to sacrifice Landfill and Emission
objective for Cost. DM3 shows some awareness about the possible outcomes for the MSW
management problem and starts from a reasonable reference point which he hoped to minimizing
cost and give an adequate value for Emission and Landfill. However, the objective values
obtained at the first iteration are not good for Cost value as he expected. Thus, he tried to reduce
the Cost value by increasing the RP for other two objectives. This kind of behaviour continues
until iteration 4, which provides an acceptable solution for DM3.

The Step A of the DM procedure was finished when all DMs obtained their optimal solution.
Step B has started with an agreement of 3 DMs, that they will use mean value of obtained RP of
each objective value for the first interaction in step B. However, the result of the model favoured
the Emission (normalized value = 0.09) objective while giving a higher value for Landfill

148
objective (normalized value = 0.38) than expected. The Cost objective was not changed much
from the RP given.

Table 6.12 Interactions of decision-making process with Multi-objective model

Reference points Normalized objective Optimal objective value

DM Time Cost Emission Landfill Cost Emission Landfill Cost Emission Landfill
1 2 3 1 2 3 (US$/d) (kg/d) (t/d)
Step A: Individual learning and evaluating the system with the model, choosing the best solution with a RP
DM1 1 0.4 0.2 0 0.68 0.14 0.28 2590.76 34886.32 20.78
Selected RP1 2 0.6 0.3 0 0.66 0.35 0.06 2541.33 38586.44 8.60
DM2 1 0 0 0 0.42 0.1 0.42 2052.13 34196.72 27.96
2 0.3 0 0.3 0.40 0.05 0.40 2016.04 33354.83 27.00
Selected RP2 3 0.4 0 0.4 0.42 0.02 0.40 2055.97 32743.67 27.29
DM3 1 0 0.3 0.3 0.27 0.34 0.57 1766.10 38370.05 36.53
2 0 0.4 0.6 0.19 0.59 0.75 1604.89 42916.83 45.79
3 0.1 0.5 0.6 0.22 0.55 0.72 1649.93 42113.38 44.24
Selected RP3 4 0.1 0.6 0.6 0.14 0.48 0.64 1489.42 40868.86 39.92
Step B: Discussion, to get the compromise solution with a compromise RP
Mean 0.37 0.3 0.33 0.42 0.09 0.38 2061.23 33957.00 26.04
Compromise 0.4 0.3 0.2 0.48 0.15 0.28 2183.20 35013.7 20.64
Figure 6.9 describes the behaviour of the model with all chosen RP. The compromise
solution seems to balance all three objectives and expected of all DMs. All proposed solutions of
DMs are the result of optimization processes, and all respect the model constraints. The final
discussion is more about solutions with real situation basis which is HAC needs a reliable
solution to reduce waste-to-landfill and emission for environmental protection as well as tourism
development. In the end, three DM agreed to choose a set of RP for 3 objectives having close
value with the mean value. it is better to spend more money because the difference in the amount
of money spent is not very high (about 100US$ per day) but can reduce 6 ton of waste to landfill,
which is important for Hoi An. As a result, a final RP was decided to obtain a result which
slightly increases the Cost and Emission to reduce the Landfill. All DMs seem satisfied with the
compromise solution and the decision-making process stopped. Table 6.12 shows the results of
decision-making procedures with all interactions of DMs with the model.

149
 Figure 6.9 MOMD model with all reference points

Figure 6.10 shows 10 obtained Pareto-optimal solution of multi-objective optimization

model of the case study with 3 DMs and visualize the procedure of decision-making process of 3
DMs step by step to reach the compromise solution.

Figure 6.10 Pareto-Optimal solutions of decision-making process

150
6.5.3.2 Solutions and decision variables

Figure 6.11 compares Cost, Emission, Landfill, GHG emission and treatment alternatives
used in for the system of the Pareto-optimal solution of each DM. The optimal solution of DM3
emitted the largest amount of GHG of approximately 700 ton CO 2-eq /d and total pollutants
emission of about 41 t/d due to a significant amount of MSW being landfill (40 t/d). However,
the cost of the solution chosen by DM3 was the most economical of approximately 1490 US$/d
because of the application of low-cost technologies (Landfill and Composting). Although the
DM2 was focused more on emission reduction, but the Pareto-optimal solution was closest to the
compromise settlement in all aspects. The solution favours a combination of AD treatment for
degradable waste and Composting treatment for sludge residual of AD and landfill for non-
degradable waste. Especially, Incineration and home treatment alternatives were not selected in
DM2’s solution. This might be due to the high emission rate of Incineration and a higher rate of
pollution emission from uncontrolled home composting than Anaerobic digestion. The Pareto-
optimal solution chosen by DM1 has the highest cost because of utilization of incineration for
the desire of landfill minimization, which leads to the highest emission of CO 2 (Figure 6.12)
contributing a major part of pollution emission.

(a) Objective values and HGH emission (b) Capacity of treatment alternatives (t/d)

Figure 6.11 Objective values and treatment alternatives of four optimal solutions

151
 Figure 6.12 Pollution emission of obtained Pareto-optimal solution from DMs

Finally, the compromise Pareto-optimal solution chose to apply various treatment

alternatives including AD for degradable waste in combination with composting for sludge
treatment. The incineration with no energy recovery is used to reduce the waste to landfill, but
due to economic and environmental objective, the capacity of incineration was relatively small.
The compromise solution balances all objectives and requirement of DMs.

Figure 6.13 present the decision variables on the system waste flow theme as result of
optimal compromise solution defined by the model, the amount of pollution emission from each
treatment alternative is also presented.

152
 Figure 6.13 Material Flow of Sustainable Waste Management
System of compromise solution

153
6.5.3.3 Sensitive analysis

The table 6.13 present the rate of waste separation at source for each optimal solutions
chosen by DMs, calculated from the result of the model. For sensitive analysis, the author will
recalculate the solution with the additional constraint of higher separation efficiency of the
system.

Table 6.13 Waste separation rate of optimal solutions

Separation rate DM1 DM2 DM3 Compromise solution

Degradable waste 97.9% 100% 67.96% 100%
Non-degradable waste 86.16% 100% 100% 90.8%
Average 92.3% 100% 84% 95.4%

The waste separation rate effects to model behaviour. The figure 6.14 shows that lower
waste separation rate results in further the distance of obtained optimal solution and reference
points. The black triangular points present the compromise reference points (0.4, 0.3, 0.2) and
other colours present the normalized objectives (  j ) obtained by different waste separation rate.

Figure 6.14 Model behaviour under uncertainty of waste separation rate

154
 Figure 6.15 Comparision of different separation efficiency
Waste separation efficiency
(%)

Figure 6.16 Distance and objective values associated with waste separation rate

155
 Figure 6.16d. shows the standard deviation (SD) of  j and reference points  j for 31

separation efficiency from 70% to 100%. The results indicate that the distance of waste
separation affects significantly to the behaviour of the model. The Higher rate of waste
separation, the closer the optimal results to the reference points. The Figure 6.16a, 6.16b and
6.16c show the real values obtained for Cost, Emission and Landfill objective associated with the
separation rate from 70%-100%. The range of separation rate show the lowest objective values is
from 92 to 96%.

6.6 Conclusion

Authors developed optimisation models for developing sustainable waste management in

HAC.

The single objective optimisation model help authority to evaluate the ability to implement
landfill reduction targets of the city. Higher landfill reduction target leads to higher cost for the
system due to the utilisation of incineration. The system cost is also influenced by the efficiency
of waste separation at source, lower rate of separation costs more. Higher landfill reduction
target also causes higher pollution emission as the result of the model. In order to reach the goal
of waste-to-landfill, the amount of waste incinerated should be increased, which can lead to the
higher emission of CO2, the largest contributor of waste management emission (Fig 6.3, 6.4, 6.5).
However, regarding GHG emission, the SC9, SC10 and SC11 emitted less GHG due to high
landfill reduction goal. Thus, incineration is an important treatment alternative for landfill
decrease and GHG emission reduction. However, in term of total amount of pollution emission,
incineration is not favoured due to a high amount of CO2 emission. Composting is the most
favoured treatment practice for degradable waste as the result of the single objective model due
to the low cost and low emission factors. However, the single objective model was not able to
evaluate the social aspects in consideration of sustainable development.

The consensus analysis model shows that minimizing cost, emission and landfill are the
most important targets of waste management in Hoi An. Also, seven treatment alternatives
including COM, AD, INC, INCE, HT, IFR and LF were chosen by local experts as potentially
suitable treatment options for the city. The multi-objective optimization model was developed
based on targets and options chosen by local experts to minimize cost, emission and landfill. The
156
result indicates that the compromise solution for sustainable waste management in Hoi An chose
the combination of AD and composting for degradable waste treatment (Figure 6.13).
Incineration is not favored when consider emission reduction. The capacity for waste incinerator
is less than 20 ton per day in all optimal solution chosen by DMs (Figure 6.11). Incineration for
energy production is no chosen due to the high capital cost.

The RP method can support citizens who generated waste involving in decision making
process. It shows a high potential in getting agreement of various DMs and reasoning conflicting
objectives. Also, RP mothed in decision making process optimize social acceptance objective of
the system since it comes to the compromise solution based on intensive discussion and
agreement of DMs.

Sensitivity analysis indicates that the waste separation at source has significant impact to
sustainable waste management. The waste management system will have more benefit (lower
cost and lower emission) with higher waste separation efficiency.

6.7 Reference

Andersen, J. K., Boldrin, A., Christensen, T. H., & Scheutz, C. (2010). Mass balances and life-
cycle inventory for a garden waste windrow composting plant (Aarhus, Denmark). Waste
Management & Research, 28(11), 1010-1020. doi:doi:10.1177/0734242X09360216
Andersen, J. K., Boldrin, A., Christensen, T. H., & Scheutz, C. (2011). Mass balances and life
cycle inventory of home composting of organic waste. Waste Management, 31(9–10), 1934-1942.
doi:http://dx.doi.org/10.1016/j.wasman.2011.05.004
Chang, N.-B., & Wang, S. F. (1996). Solid Waste Management System Analysis by
Multiobjective Mixed Integer Programming Model. Journal of Environmental Management,
48(1), 17-43.
Chang, Y. H., & Chang, N.-B. (1998). Optimization analysis for the development of short-
team solid waste management strategies using presorting process prior to incinerators. Resources,
Conservation and Recycling, 24(1), 7-32. doi:http://dx.doi.org/10.1016/S0921-3449(98)00036-6
Chiu, C.-T., Hsu, T.-H., & Yang, W.-F. (2008). Life cycle assessment on using recycled
materials for rehabilitating asphalt pavements. Resources, Conservation and Recycling, 52(3),
545-556. doi:http://dx.doi.org/10.1016/j.resconrec.2007.07.001
Chu, M. T. (2014). Báo cáo tổng kết đề tài: Xây dựng mô hình đồng quản lý rác thải tại hai xã,
phường Cẩm Hà và Cẩm Phô, thành phố Hội An (Project report: building co-operative waste
management model in two wards, Cam Ha and Cam Pho, Hoi An city). Retrieved from Hoi An,
Viet Nam:

157
 Costi, P., Minciardi, R., Robba, M., Rovatti, M., & Sacile, R. (2004). An environmentally
sustainable decision model for urban solid waste management. Waste Management, 24(3), 277-
295.
Decision:31/2014/QĐ-TTg. (2014). on support mechanism for the development of power
generation projects using solid waste(s) in Vietnam Ha Noi, Viet Nam Retrieved from
http://gizenergy.org.vn/media/app/media/PDF-Docs/Legal-Documents/Solid-
Wastes/Decision%2031-2014-QD-
TTg%20on%20support%20mechanism%20for%20SW_EN.pdf.
DEFRA. (2004). Review of Environmental and Health Effects of Waste Management:
Municipal Solid Waste and Similar Wastes. Retrieved from
https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/69391/pb9052a-
health-report-040325.pdf
EPEM. Database of Waste Management Technologies. AD 5: Complete Dry AD (Dry
Fermentation). Retrieved from http://www.epem.gr/waste-c-control/database/html/AD-05.htm
FHWA. (2016). Report FHWA-RD-97-148: User Guidelines for Waste and Byproduct
Materials in Pavement Construction. Federal Highway Administration Research and Technology.
Fiorucci, P., Minciardi, R., Robba, M., & Sacile, R. (2003). Solid waste management in urban
areas: Development and application of a decision support system. Resources, Conservation and
Recycling, 37(4), 301-328. doi:http://dx.doi.org/10.1016/S0921-3449(02)00076-9
Galante, G., Aiello, G., Enea, M., & Panascia, E. (2010). A multi-objective approach to solid
waste management. Waste Management, 30(8–9), 1720-1728.
doi:http://dx.doi.org/10.1016/j.wasman.2010.01.039
Harijani, A. M., Mansour, S., & Karimi, B. (2017). A multi-objective model for sustainable
recycling of municipal solid waste. Waste Management & Research, 35(4), 387-399.
doi:doi:10.1177/0734242X17693685
Hung, M.-L., Ma, H.-w., & Yang, W.-F. (2007). A novel sustainable decision making model
for municipal solid waste management. Waste Management, 27(2), 209-219.
doi:http://dx.doi.org/10.1016/j.wasman.2006.01.008
IPCC. (2006). 2006 IPCC Guidelines for National Greenhouse Gas Inventories - Vol 5: Waste.
Institute for Global Environmental Strategies (IGES), Hayama, Japan: IPCC National
Greenhouse Gas Inventories Programme.
Khan, M. R., Mcmahon, M., & Decanio, S. (1991). Sewage sludge: a fascinating feedstock for
clean energy. Paper presented at the Symposium on Clean Energy From Waste and Coal Energy
From Sewage Sludge, Biomass and Municipal Waste.
Khoo, H. H., Lim, T. Z., & Tan, R. B. H. (2010). Food waste conversion options in Singapore:
Environmental impacts based on an LCA perspective. Science of The Total Environment, 408(6),
1367-1373. doi:http://dx.doi.org/10.1016/j.scitotenv.2009.10.072
Kim, M.-H., & Kim, J.-W. (2010). Comparison through a LCA evaluation analysis of food
waste disposal options from the perspective of global warming and resource recovery. Science of

158
The Total Environment, 408(19), 3998-4006.
doi:http://dx.doi.org/10.1016/j.scitotenv.2010.04.049
Lee, S.-H., Choi, K.-I., Osako, M., & Dong, J.-I. (2007). Evaluation of environmental burdens
caused by changes of food waste management systems in Seoul, Korea. Science of The Total
Environment, 387(1), 42-53. doi:http://dx.doi.org/10.1016/j.scitotenv.2007.06.037
Lyeme, H. A., Mushi, A., & Nkansah-Gyekye, Y. (2017). Review of multi-objective
optimization models for solid waste management systems with environmental considerations.
Journal of Mathematical and Computational Science, 7(1), 150.
MOC. (2012). Decree 322/QĐ-BXD: Về việc công bố suất vốn đầu tư xây dựng và mức chi
phí xử lý chất thải rắn sinh hoạt (announcement of Investment rate and operation cost level for
municipal solid waste treatment)(In Vietnamese). Ha Noi, Viet Nam Retrieved from
http://moc.gov.vn/en/web/guest/home/-/legal/2pBh/vi_VN/18/150170/37.
Ngoc, N. (2016) Personal interview on Novermber 15, 2016/Interviewer: H. M. Giang. Hoi
An Public Work Service Ltd.
Solymosi, T., & Dombi, J. (1986). A method for determining the weights of criteria: The
centralized weights. European Journal of Operational Research, 26(1), 35-41.
doi:http://dx.doi.org/10.1016/0377-2217(86)90157-8
Tan, R. B. H., & Khoo, H. H. (2006). Impact Assessment of Waste Management Options in
Singapore. Journal of the Air & Waste Management Association, 56(3), 244-254.
doi:10.1080/10473289.2006.10464463
Zimmermann, H.-J. (2011). Fuzzy set theory—and its applications: Springer Science &
Business Media.

159
7. CONCLUSION AND RECOMMENDATION

The aim of this dissertation was to develop a methodology of solid waste management
system planning using MODM towards environmental efficient, economic affordability and
social acceptance. The case study was conducted in Hoi An, a famous tourist city in Viet Nam. A
holistic method approach for developing a waste management plan was discussed step by step in
detail. Results were analysed and discussed to explain the model. Also, some concerning
contents were provided, such as waste generation relevant households factors and improvement
for household waste collection system or an increase of fee for business sectors. The evaluation
of landfill reduction alternatives and waste separation rates influences to the waste management
cost and system emission by single optimisation model was provided. Finally, MODM model
and decision-making process by RP method suggested a sustainable solution for waste
management in Hoi an city. The optimal solutions provided a detailed flow of waste materials,
appropriate treatment technology and their size.

7.1 Summary of key findings

An overview of the research area was introduced in the first section. Rapid urbanisation and
industrialisation in Viet Nam caused the increase of waste generation and the variety of
composition. In addition, inappropriate waste management system in Viet Nam has led to
various environmental and health issues. Sustainable solutions for waste treatment and waste
stream management is essential to improve the current status of waste management in
Vietnamese cities. A study on sustainable waste management using MODM model has never
been carried out in Viet Nam. Thus the study proposed a MODM model to define optimal
solutions of sustainable waste management in Viet Nam.

Through the review of existing literature regarding MODM model as well waste
composition and generation studies. The model proposed in this dissertation tends to cover some
gaps of previous MODM model studies in waste management including optimization of social
acceptance objective, measurement more pollutants emission for environmental objectives and
consideration of waste cycle. To gather the input data and parameters for the model's various
methods were applied such as a sampling waste at source for waste composition and generation,

160
face-to-face interview for collecting household’s socioeconomic information and laboratory for
waste characteristics and calorific values. Also, secondary data were collected from Hoi An
waste treatment plant to estimate daily waste transported for treatment and local authorities’
offices for demographic and economic information. Moreover, in considering waste cycle and
multi-pollutants emission in the model, reliable studies and scientific reports in LCA and LCI
were referred. To identify the principal objectives of the waste management system and potential
treatment alternatives in the city, a face-to-face interview with 18 local experts was conducted.
The main achievement is presented as follows:

1. Daily per capita household waste generation was 0.223 kg capita -1 day-1. Total household
waste generation rate in Hoi An City was 20.75 tonnes per day, which accounted for
about 30% of total municipal waste generation. People living in rural areas generated
about half of the amount of daily waste compared to people living in urban areas. On the
other hand, residents in the urban areas highly affected by tourism activities produced
less waste than in areas not affected by such activities. Hotels generated about 0.6 kg
room-1 day-1 and one restaurant in HAC produced an average of 26.18 kg day-1. The
mean of waste generation from tourist streets was 6.99 kg 100m -1 day-1. Regarding the
correlations between the waste generation rate and the number of hotel rooms, the result
indicated a statistically significant correlation between the number of rooms and total
waste generation of hotels.

2. The municipal waste composition of the city had food waste as the largest proportion
(42%) and hazardous waste as the smallest contributor (less than 1%). Total
biodegradable waste (food and yard trimmings) was approximately 53%. Combustible
waste was the second significant component of about 16% while other recyclable
contributed about 20% of the municipal waste composition. Waste composition in the
city centre was affected by tourism services because it was different from other areas
and likely alike with the waste composition of the tourism sector in food, garden, and
some other components. Results of ANOVA shows that waste generation is significantly
different in urban and rural areas. Meanwhile, there is no significant difference in waste
composition among three strata.

161
 3. Recyclable waste (plastic, cardboard, paper, and metal) accounted for a significant
percentage of around 20% giving a good chance for material recovery. However,
informal recycling activities for plastics, cardboard, and some other materials by
organized waste pickers at treatment plants are currently adequate. Also, current
recycling activities employ residents living nearby Hoi An waste treatment plant and are
supported by local people. A material recovery program, therefore, may not be
warranted for now because of social and technological constraints but should be taken
into account in the future.

4. Results of this study suggested a methodology of waste generation and composition

survey, the classification category list for sustainable waste modelling and planning. The
results and methodology are useful for decision makers, authorities, planners as well as
researchers to develop or modelling sustainable waste management.

The results of waste physical and chemical characteristics also analysed and discussed:

5. The moisture content of MSW in Hoi An is about 53%. The high moisture content is
caused by Food and Garden waste, which accounted for 40% and 10% in composition
and the moisture content is 75% and 57% respectively. The high moisture will affect the
thermal treatment of MSW but is essential for biological treatment technology. Also, the
C:N ratio of biodegradable waste (Food and Garden waste) is 22.1, which is a good ratio
for composting. However, since the composting planting in Hoi An has been operating
ineffectively, a deeper study in socio-economical and technical efficient assessment
might be needed to improve the treatment efficiency.

6. The average heating value of MSW is about 9.4 MJ/kg waste (ranged from 2 – 30 MJ/kg).
An incineration could be a potential technology to reduce the waste amount quickly in
the city. WTE has been one of the best technology in term of energy recovery in MSW
management. The result from energy content of MSW in Hoi An indicated that a WTE
system might be a potential option for the city waste treatment system.

Another content included in this study is household relevant factors of waste generation and
waste generation predictive model were presented with following findings:

162
7. The study indicated that the impacts of socio-demographic, geographic variables and
family economic activities on waste generation rate are highly significant. The location
of the house is the most affected predictor for the daily amount of waste generated per
capita. The model showed that a person in an urban house produced much higher solid
waste (70-80%) than one living in a countryside family. Another important factor
influenced the household waste generation in HAC is family economic activities. If a
family is holding a business at home, the waste generation rate of the household will
increase by about 35%. The study found that household size and the area of the house
are also significant determinants of per capita waste generation while other variables,
particularly income per person, prove to not significant as correlates of waste production.

8. Regarding waste generation modelling, results from this study demonstrate that Bayesian
Model Average method was a robust method to determine solid options in multiple
linear regression models, especially dealing with a big number of independent variables.
The result of BMA indicated that the linear regression model with four independent
variables was the best model for estimating waste generation in HAC. Meanwhile, the
model with three regressors has slightly lower performance but is also useful to predict
the waste generation of household quickly because of availability information of
predictor variables in census database. Linear tests and model validation results proved
that both model a reliable for estimating waste generation in the city. The study
attempted to increase the understanding of waste generation to support waste
management planning for the city. Thus, the two models are useful in addressing waste
generation for not only analysing key factors influencing the waste generation, but they
also giving waste managers a link to have a model for estimating waste generation to
improve waste reduction and management efforts. The results and methodology are
accepted to be informative for authorities, decision makers, stakeholders and planners to
develop a waste management plan.

9. The multivariate linear regression model provides an exact estimation of waste

generation in rural and urban areas in better calculation and arrangement of vehicles,
labours, collection routes. It also suggests that a decentralised treatment approach could

163
 benefit to reduce the collection cost for the small amount of waste generation in further
areas. However, the approach needs to be studied carefully to have the applicable
implications and appropriate legislation to encourage decentralised waste treatment.
Also, the Waste fee for business sector might be an important factor to take into
consideration in waste collection and management planning when the number of
household opening business at home for tourism and local services such as small
restaurants, homestays, shops, convenience stores and vehicle rental shops have been
rising gradually in the city.

To evaluate the ability of landfill reduction implementation and effects of waste separation
efficiency, a single optimization model minimizing the total cost of the system were developed
under various possible analysis scenarios. The model was searching for the optimal solution of
minimizing cost for 12 examined scenarios, created by the combination of three values of landfill
target (50%, 25% and 10%) and four values of separation rate (100%, 90%, 80% and 70%).
Multiple assessments based on main aspects such as pollution emission (including CH 4, CO2,
NOx, SO2, N2O, VOC, heavy metal…), GHG emission, cost, land use, etc. The key findings are
shown as follows:

10. The Cost is proportional with the landfill target, the higher landfill reduction target the
higher cost for the system. The target of 10% waste-to-landfill cannot be reached if the
rate of waste separation at source is lower than 70%. For each landfill target, the lowest
cost is found when MSW is separated completely (SC1, SC5 and SC9) which has the
efficiency of waste separation of 100%. Thus, the separation efficiency affects
significantly to the system cost. The lower separation efficiency leads to higher cost of
the system. The waste separation has the same effect on emission, a lower rate of waste
sorted at source causes higher emission (comparing among the scenarios with the same
LT (SC1-SC4: 50%, SC5-SC8:25% and SC9-SC11:10%).

11. Incineration is an important treatment alternative for landfill and GHG emission
reduction. However, in term of total amount of pollution emission, incineration is not
favoured due to a high amount of CO2 emission. Composting is the most favoured
treatment practice for degradable waste as the result of the single objective model due to

164
 the low cost and low emission factors. However, the single objective model was not able
to evaluate the social aspects in consideration of sustainable development.

Multi-objective model and decision making procedure were applied to identify sustainable
solution for waste management in Hoi An. Firstly, the objectives and potential treatment
alternatives were identified via a face-to-face interview questionnaire survey, analysed by
consensus analysis model. Then, the multi-objective model was developed with three objective
functions including minimizing cost, minimizing emission and minimizing waste to landfill with
seven potential treatment alternatives. The decision-making process was conducted with three
decision makers presenting authority, citizen and waste management contractor using RP method.
A sensitivity analysis was carried out to evaluate the model behaviour as well as the objectives
value due to the variation of waste separation rate. The major findings are presented as follows:

12. The result indicates that the compromise solution for sustainable waste management in
Hoi An. The capacity for the waste incinerator is less than 20 ton per day in all optimal
solution chosen by DMs. The compromise solution showed that the total cost of the
system is about 2200 $/d, the total emission of various pollutant is 35 t/d and waste to
landfill is approximately 20 t/d. GHG emission of chosen solution is about 3.8 t/d.
Incineration for energy production is not chosen due to the high capital cost. The
combination of AD and composting was suggested for degradable waste treatment, in
which degradable waste is processed in AD then the sludge residual will be treated with
composting. Incineration is not an optimal option when considering emission reduction.
Thus, the amount of waste incinerated is relatively low. The results indicated that
collecting and treating degradable waste separately is a sustainable option for HAC (also
other cities in Viet Nam) due to high moisture, low calorific values of food waste as well
as low cost and low emission factors of treatment technologies (anaerobic digestion and
composting). Incineration contributes significantly to the reduction of waste to landfill,
but a high financial burden and CO2 emission.

13. The RP method and CAM can support citizens who generated waste involving in the
decision-making process, to optimizing the social aspects in optimization model. It
shows a high ability in supporting DMs for agreement and reasoning conflicting

165
 objectives. Also, RP method in decision-making process optimizes social acceptance
objective of the system since it comes to a compromise solution based on intensive
discussion and agreement of DMs. Sensitivity analysis indicates that the waste
separation at source has a significant impact on sustainable waste management. The
waste management system will have more benefit (lower cost and lower emission) with
higher waste separation efficiency.

7.2 Limitations of the study and recommendations for future studies

The dissertation aimed at to develop a holistic methodology of sustainable waste

management planning for cities in Vietnam by using MODM model. However, there are
limitations and lessons learned from this study.

1. Regarding waste generation and composition, household waste and tourist waste
generation was well and detailed identified but not other sources such as hospital,
institutional, industrial, market waste. Further studies should focus on waste from more
sources.

2. In addition, tourist waste has a major contribution in the city as well as great variation in
different types of source. Thus, detailed study in tourist waste generation and
composition based on scales of sources, characteristics of sources and location of
sources will provide beneficial data in model calculation. Studying on tourist waste
generation and composition is being carried by another member of our lab.

3. The seasonal variation was not concerned in this study. Thus, solid waste characterization
variation between dried and rainy season was not verified. Waste characterization
associated with seasonal variation study is also recommended.

4. The linear regression models constructed in the current study is valuable in estimating
waste generation as it provides observational evidence of the influences of multiple
factors. However, it is not reliable for predicting waste generation in the future. Lack of
historical data in caused difficulties for modelling waste model predicting the future.
Thus, modelling municipal waste generation model which can forecast the future is
essential and needed to be studied. Consumption habits and attributions of residences

166
 should be studied in association with waste generation and composition, which can
provide additional information in predicting waste generation and composition.

5. The parameters for waste flow, waste residual, pollutant emissions factors in the model
proposed were referred from other studies. The author was not able to conduct
independent LCA and LCI study for various waste treatment process to obtain the
parameters due to the complexity, high cost and time-consuming. Thus, LCA and LCI
studies for different waste treatment technologies in Viet Nam is essential. That will
contributes significantly in waste cycle calculation and model precision and therefore,
provide more accurate results.

6. The study proposed a deterministic model to identify the optimal solution for the city.
The sensitivity analyses were conducted in this study, but just for the variation of waste
separation efficiency, other variations such as waste generation input, variation in cost
was not able to be conducted. Thus, it is recommended that more sensitivity analysis
with more variation of parameters should be assessed to examine the diverse of the
solutions. Stochastic modelling of the system is also suggested to identify the change of
the solution when waste generation input varies.

167
 ACKNOWLEDGEMENT

I would like to express my sincere gratitude to my principal supervisor Prof. Fujiwara

Takeshi for the continuous support of my Ph.D study and related research, for his patience,
motivation, and immense knowledge. His valuable advices, creative ideas and excellent guidance
helped me in all the time of research and writing of this dissertation. Thanks for his enthusiasm
and patience, not only in teaching me to do research, but also in instructing me the Japanese
cultures and working manners, I improved myself in many ways. I could not have imagined
having a better advisor and mentor for my Ph.D study.

Besides my advisor, I would like to send my profound gratitude to Prof. Kawamoto Katsuya
and Prof. Matsui Yasuhiro, who gave me insightful comments and enthusiastic supports, and also
excellent suggestions which incented me to widen my research from various perspectives as well
as my dissertation composition.

My sincere thanks also goes to Ms. Chujo Yoko, who provided me kind support and great
encouragement throughout the period of my study, and Ms. Ono Yoko, a very nice, friendly and
helpful person who has been constantly supportive during my first two years. Thank you for your
supports and guidance to help me in life and work since the first day I have been in Japan.
Without your precious support it would not be possible to conduct this research.

I also wish to convey my grateful to Prof. Watanabe Masaji for his great mathematical
modelling lectures that I attended, also his patience and enthusiastic suggestions.

I would like to thank people in Sanyo Bigyo Co. Ltd, where I stayed during my internship.
Special thanks to Mr. Muroyama Koichi, who gave me great support and helpful guidance in the
company.

I am grateful all the people that I worked with in Hoi An city, Mr. Quang (Hoi An
Environment and Natural Resource Office), who provided me a lot of important information; Ms.
Ngoc and Ms. Thuy (Hoi An Public Work Service Ltd. Co), who helped me a lot when I worked
in the field. Also, I would like to thank to director board of Hoi An Puplic Work Service Ltd. Co
and College of Technology - The University of Danang for the use of their facilities. The

168
financial support from the Research Grant for Encouragement of Students of Okayama
University is also greatly acknowledged.

I thank to my friend Dr. Nguyen Phuc Thanh for his long and informative discussion also his
excellent experience sharing. I thank to a fellow, Mr. Dinh Tien Tai for his interesting discussion
and his helpful comments in data analysis. I also thank to my collegues in Department of
Environmental Technology and Management, National University of Civil Engineering in Viet
Nam, who always support and cheer me up. Special thanks to Prof. Nguyen Thi Kim Thai, who
provided me huge support and valuable suggestions in my research.

I thank my fellow labmates in for the stimulating discussions, for the funny time we were
spending and working together. My sincere appreciation goes to Mei, Seng, Lena, Dinh, Yoneda,
Wada, Ogasawa, Fukuyama, Yamashita, and other members for your kindly help, your support
and providing favorable conditions for my study in the lab and for all the fun we have had in the
last three years. Special thank is due to Toan, my lab-mate for his valuable support, constant and
continuous help during the time of conducting survey and sampling in Hoi An city. I also thank
to Phuoc, who help me a lot in analyzing waste samples.

I also thank my friends (too many to list here but you know who you are!) for providing
support, friendship and love that I needed. I would like to thank to all of you for sharing,
enjoying with me, always being supportive throughout my time here and for enrich my life in
Japan in many ways.

I especially thank my mom, dad, and sister, brother-in-law and my niece, who gave my grate
support and love. My hard-working parents provided me unconditional love and care. I would
not have made it this far without them and I know I always have my family to count on when
times are rough.

I cherish the inspiration of my best friend, soul-mate, and beloved wife. I married the best
person out there for me. Phuong is the only person who can appreciate my personality and sense
of humor. She has been a true and great supporter and love me unconditionally during my good
and bad times. These past three years have not been an easy ride, both academically and
personally. I truly thank her for sticking by my side, even when I was irritable and depressed in
struggling for research. I feel that what we both learned a lot about life and to live life to the
169
fullest. There are no words to convey how much I love her. Also thanks to my beloved son, Minh
Duong, my treasure, energy recharger and mental power for his unconditional love. This work
cannot be finished without their love and support.

170